BabyNest

[warm bay]

Can you elaborate

[stoic igloo]

name
email
baby name
date_of_birth
expected_due_date
fundal height or waist size with date [{}]
weights with date [{}]
other health stuff,

[warm bay]

Do we have to store that data to. I thought earlier you said we just need location and due date

[stoic igloo]

we can store this info in device

and also pass it to llm for better results

[warm bay]

Ok so I'll do some searches and share what data needs to be shared

[warm bay]

That'll be better ig

[stoic igloo]

key-value pair in local db

[warm bay]

Ya now I understood

What do you think about this @stoic igloo?

Like a dataset for testing the models. That way we can compare accuracy in much better and quantitative way

[dry meadow]

yes we are almost completed with the integration will update asap

[dry meadow]

@stoic igloo when can we have a meet regarding the ios issues

[boreal smelt]

I have an ios device I can help with the issues

[dry meadow]

https://github.com/AOSSIE-Org/BabyNest/pull/31

this PR is in continuation with the setup sqlite for backend PR
@stoic igloo Please have a Look

[cinder mural]

@stoic igloo Pls review, I have solved the issue #7 https://github.com/AOSSIE-Org/BabyNest/pull/32

[warm bay]

Hey @cinder mural do you have ios device?

[cinder mural]

Yes, I have ios device and macos

[dry meadow]

that's great can you please help us resolve the navbar issues

[cinder mural]

I am on it, trying to solve the whole

[dry meadow]

we can get on code?

call*?

[cinder mural]

Yes, we can rn ?

[dry meadow]

um give me some time
lets get @warm bay availability also

[dry meadow]

but we can schedule one tonight if you are free?

[cinder mural]

Yes ofcource

Update this here I will make myself free

[dry meadow]

Sure @10 would be good for you?

[dry meadow]

done pls have a Look

[warm bay]

10 works for me

[cinder mural]

Can you make it 11:30 or onwards tonight?

Or it is already done ?

[dry meadow]

Sure

We can do it little late

I was also outside thanks for Reminding

[cinder mural]

No problem

[dry meadow]

Can we do now?

[cinder mural]

Okay getting rdy in 5 min

[dry meadow]

Sure

Can we do on Google meet?

No

Ios issues

[cinder mural]

Yes ofc

Share me the link

[cinder mural]

Link

[dry meadow]

check dm

[warm bay]

https://github.com/AOSSIE-Org/BabyNest/pull/35

Hey @stoic igloo , I have managed to run a local llm of size less than 500MB smoothly. Its accuracy is pretty good. Would share some accuracy scores based on datasets asap.

Further if any model needs to be tested in future, we just have to change some variables in .env and we're good to go. This code makes sure that everything run smoothly without disturbing any other functionality.

Please have a look @stoic igloo

[dry meadow]

Yes we are using a model of 0.5B params

[worthy citrus]

@dry meadow You are now level 22! :confetti_ball:

[dry meadow]

Size can be further reduced..., other models can be integrated as well

[cinder mural]

@dry meadow let’s connect tonight if you are free, I find a way to fix the issue we talked yesterday

@warm bay @dry meadow confirm me

[dry meadow]

Sure let's do it today

[cinder mural]

I have tried to solve issue #20 if it also works properly in android it is good, either I will make this changes platform specific
https://github.com/AOSSIE-Org/BabyNest/pull/36
@stoic igloo @dry meadow

[dry meadow]

oh so you have not checked for android?

[boreal smelt]

@warm bay tested this llm it gives great results to the queries.

[cinder mural]

No I am just asking for review

[worthy citrus]

@cinder mural You are now level 3! :confetti_ball:

[boreal smelt]

@dry meadow can we have a smaller sized model which gives better results?

[warm bay]

It would be better if we try to reduce this model's size only

I am working on it

[lyric berry]

great, qwen models work awsm.
can help in quantization/optimization part...

[dry meadow]

like we are currently making it work on 450mb so do we need to reduce size atp

?

we already reached our benchmarked size limit so we can now focus on accuracy part

[lyric berry]

ahh no ig the cap was around 400 MB right, this too would be good. but model size can still be decreased

[warm bay]

Yes, I'm currently working on it, Just researching about other techniques to reduce the size

[dry meadow]

ya 400mb-500mb

[lyric berry]

try using PEFT techniques, would be efficient and fast

[worthy citrus]

@lyric berry You are now level 9! :confetti_ball:

[stoic igloo]

that's great

[warm bay]

Yup, raised a pr for that too

Please have a look

[stoic igloo]

yup

[warm bay]

Yes

[dry meadow]

Yes we used GGUF file to do this

[warm bay]

The pr was in collaboration with ishaan

[dry meadow]

took time to understand the mechanism but were able to do this

[warm bay]

But everyone has done great work in the research

[dry meadow]

further if we want to increase accuracy can use a more params model around 1B or 1.7B

but if we want size to be less as possible then this is good 0.5B params

[warm bay]

@lapis field @terse elbow @lyric berry @spare cradle

[dry meadow]

yes everyone's suggestions was helpful

[warm bay]

Ya we'll do as you say

[dry meadow]

faced many issues bringing this to life @warm bay but in the end were able to do

[warm bay]

But I think the current accuracy is pretty good

[dry meadow]

Ya current accuracy is pretty good

[warm bay]

Maybe little more prompt engineering and we're good to go

[dry meadow]

will be raising a PR for this

[warm bay]

Yup

[stoic igloo]

@warm bay can you test this in android?

are we good to go?

[dry meadow]

Ok I can test

[cinder mural]

Yes we should before moving further

[warm bay]

https://github.com/AOSSIE-Org/BabyNest/pull/31

[stoic igloo]

there are some PR issues guys

[warm bay]

Resolved the conflicts

Ready too merge

[stoic igloo]

will test it

[warm bay]

Sure

[dry meadow]

yea resolved the conflicts

[stoic igloo]

remove the files not required

[dry meadow]

actually I was testing this for customzing the whole UI theme

from the settings

[stoic igloo]

alr

[dry meadow]

Yes will raise a PR for the customize option asap

[stoic igloo]

remove the comments

[dry meadow]

sure

[stoic igloo]

keep the code as clean as possible

avoid including files like package-lock.json in your commit

[dry meadow]

removed these

[dry meadow]

is this file in PR SR No. 2?

not able to track

[dry meadow]

Sure will keep this in mind, actually because of so many features we have to change so comment some parts for testing

But will keep these things in mind

[stoic igloo]

yeh i get it

include a video of your changes

[dry meadow]

Yes attaching

[stoic igloo]

include in the next PR, remove comment
and make the code cleaner

and as small as possible
using functions

@warm bay
conflicts in 35

[dry meadow]

Ok got it

Now will work on making the code cleaner and more optimised

just wanted to make the basic setup

[dry meadow]

Resolved

[warm bay]

Resolved

[stoic igloo]

@cinder mural

[cinder mural]

Okay, I am checking

[warm bay]

Take care man

Yes currently exploring the techniques to reduce llm size

[lyric berry]

yes, PEFT ft would be good enough

[terse elbow]

me too @warm bay due to examsss being inactive

sorry yrr

[cinder mural]

@stoic igloo I have resolved the conflicts by creating new PR pls check it out https://github.com/AOSSIE-Org/BabyNest/pull/39

[narrow night]

Hello Everyone, I'm thinking of adding a Medicine Tracker feature to the BabyNest app. This would help moms keep track of their medications, set reminders, and maintain a history of taken doses.

The idea includes:

Adding, editing, and deleting medications
Customizable reminders and notifications
A log of taken and missed doses
A simple and user-friendly tracking interface
I would love to hear your thoughts. Do you think this would be useful? What else can be added or improved? @stoic igloo @proper musk https://github.com/AOSSIE-Org/BabyNest/issues/44

[rigid wedge]

@proper musk @stoic igloo

[vital turret]

Hello @stoic igloo @proper musk ,

I am Pulkit Porwal and I am a 2nd year BCA Student and I would love to contribute to this project. I am very proficient in React or React Native and I also have a lot of experience in Integrating AI for different use cases.

Can you please tell me the ideas or features we are going to build.

I am trying to run the application but getting this error @anyone please help

[cinder mural]

Try to install all the dependencies properly, the app is working properly in both android and ios

also check system setup

[brittle yoke]

Hello everyone, I'm Austine, a react and react-native developer and a graduate of ALX. I'm passionate about being part of the development of this app.

Let me go straight to the GitHub repo and fish out some issues that need to be resolved.

[brittle yoke]

@stoic igloo is there any pending task to work on for this app?

[narrow night]

Hello Everyone,
I've raised an issue about inconsistent theme colors and fonts, making text hard to read. Please take a look. Happy to improve it further and Looking forward to your suggestions also!https://github.com/AOSSIE-Org/BabyNest/issues/52 @stoic igloo @proper musk

[stoic igloo]

You can work on AI agent feature

[dry meadow]

Hi @stoic igloo, what are the plans for deployement?

are we planning to deploy this after this gsoc period?

[stoic igloo]

Yep

[dry meadow]

Sure will be including a plan for deployement in my proposal

[dry meadow]

also apart from the ai agent and customizations in UI, any additional feature you would like us to implement?

[stoic igloo]

It will be the end goal and we plan to release a smooth working version of the app by July

[worthy citrus]

@stoic igloo You are now level 29! :confetti_ball:

[dry meadow]

Sure

Also like we have achieved many of the expected outcomes which were listed on the aossie page, the agent part is left which is the main part

but apart from it what else can we add ?

like some other pages medicine tracking or anything else?

[stoic igloo]

No, for now lets focus on the making the basics
Also can we update the readme so we can easily setup the backend and backend-frontend integration as well

[dry meadow]

Sure I will update the readme for the integration part

[stoic igloo]

Yep that was a great issue raised

We can work upon that and also keep track of medicines with the agent
Also, dont forget Reminders and notifications is a very big part of our app

[dry meadow]

yes yes that I am working upon

[dry meadow]

can we also have a section for uploading the scans and ultrasound?

like i guess this idea was also being discussed by someone

[stoic igloo]

Nope, not required

[warm bay]

Bhavik can we do tracking of important parameters like weight, medicines etc

Like they matter a lot during pregnancy so that'll good for user if they can track it in a single app only

[dry meadow]

yes we can have a section for weight tracking

i guess weight has to be recorded for each week and sent to agent

[warm bay]

Yes

[narrow night]

what about medicine tracker?

[warm bay]

So my point is track all the things that affect a user during pregnancy

And then we can send all that to agent

[warm bay]

Ig we can go ahead with that

[narrow night]

okay

[warm bay]

.

Ok so I am working on reminders and notifications

Ishaan you can work on other parameters that need to be tracked

Also @stoic igloo can you please review the pending prs

[dry meadow]

Yes I am working on the parameters that need to be tracked and integration of user details with backend

[fleet vessel]

@stoic igloo please assign me this issue https://github.com/AOSSIE-Org/BabyNest/issues/53

[fleet vessel]

@everyone do we need to get assigned before working on the issue in this project ?

[worthy citrus]

@fleet vessel You are now level 8! :confetti_ball:

[warm bay]

Hey @fleet vessel Please refer to the above convo

Bruno has decided that there will be no authentication in this app

As this is intended to be a light weight go-to app

[brittle yoke]

Okay on it.

[warm bay]

That's why it will be offline and no auth is needed

[warm bay]

Hey @brittle yoke can you share what functions and parameters are you thinking to pass to the agent

And how are you planning to make it offline

[warm bay]

So ig there will be no supabase (neither for auth nor db) as of now

[fleet vessel]

okay Thank you.

[stoic igloo]

No online dbs required

[warm bay]

.

[brittle yoke]

Okay..
Here are the features I would like to implement

1. Nutrition & Wellness Recommendations
   b. AI-generated meal plans based on dietary needs and restrictions.
   c. Hydration & vitamin intake tracking.
   d. Exercise suggestions based on trimester and health conditions.

2. Appointment Reminders & Scheduling
   Feature: AI suggests appointment slots based on user availability & doctor visits.

Tools & Tech Stack:
Calendar Integration
AI Optimization
Push Notifications

Implementation Flow:
User inputs doctor’s availability.
AI suggests best appointment slots based on location, past checkups, and availability.
Sends confirmation & reminders via push notifications.

3. AI Chat Assistant for Health Guidance
   Feature: Conversational AI answering pregnancy-related questions 24/7.

Tools & Tech Stack:
AI Model: OpenAI GPT-4 / Google Vertex AI
Chat UI: react-native-gifted-chat
Speech-to-Text (Voice Assistant): expo-speech + react-native-voice
Knowledge Base: Firebase Firestore (for structured FAQ responses)

Implementation Flow:
User asks a question (“What foods should I avoid?”).
Provides response as text & voice.

You can go through it and let me know.
More would come why doing my user research.

[warm bay]

Yeah so the ai chat assistant for text has already been implemented

And for the other features, we have to use the models that can be ran offline

[dry meadow]

Yes @brittle yoke the app has to be fully offline

[brittle yoke]

Without connection right?

[dry meadow]

Ya without a network connection

[warm bay]

Yes

So the model should be small in size

Like currently the model we are using is of 400mb approx

[brittle yoke]

How about Offline Prenatal Workout Guide?

[worthy citrus]

@brittle yoke You are now level 2! :confetti_ball:

[brittle yoke]

What do you think?

It's just one offline feature for the AI agent.

Offline Blood Pressure & Heart Rate checker.

🤔🤔🤔

[warm bay]

I am not sure that we need tracking tools in the app as of now

[dry meadow]

how will we check bp and heart rate through the app? or are you talking about tracker?

[brittle yoke]

It's just like answering some questions and then boom AI analyses it and give you the result and then maybe check.

Because I have a smart watch that checks my heart rate.

[brittle yoke]

Oh?

How about you give me a breakdown of what has been done on the AI agent then from there I would know what and what extra feather will be added.

[dry meadow]

I dont think we are planning this

[dry meadow]

chat has been built

[stoic igloo]

nothing so far

[stoic igloo]

will do later

[warm bay]

Ok

[brittle yoke]

Someone said the chat has been built.

[dry meadow]

It doesn't support the agentic framework right now

[brittle yoke]

Cool.

The way it's going to work right now is if I can know the features that have been done, I think I can come up with lotta ideas for the offline section.

[warm bay]

Yeah please connect with me for that

Working for a while on this project

[brittle yoke]

Oh..both of us?

[worthy citrus]

@brittle yoke You are now level 3! :confetti_ball:

[brittle yoke]

You're a front end dev?

[warm bay]

Mostly backend and ai dev

But I do like little bit frontend too😢

[dry meadow]

Ping me also will connect

Worked on babyNest from scratch itself

[brittle yoke]

Okay.

So we all work together right?

Let's wait for how far they have gone to see our inputs.

What do you think?

I'm still doing my user research on this.

[dry meadow]

Yes definately would love to collaborate with you

What is your time zone?

[brittle yoke]

So are we to wait till we get this info before proceeding in doing a PR?

[dry meadow]

Yes we can get on with work

Agentic framework has to be built as it is an expected outcome

[warm bay]

Yup

[brittle yoke]

Okay

[dry meadow]

Hi @stoic igloo can we have a landing page as a website for the app where there will be option to download on ios and android with basic description and contact us page?

I can get on with this asap if this is required

https://github.com/AOSSIE-Org/BabyNest/issues/57

raised a feature request for this

How does this idea look to everyone? @brittle yoke @warm bay @narrow night

[brittle yoke]

Just like this right ?

https://moniepoint.com/ng/business

Scroll down

[narrow night]

@dry meadow This sounds like a great idea! A landing page will make it easier for users to access and learn more about the app.

[brittle yoke]

Yessss

Then you go straight to download the app both on IOS and Android.

[dry meadow]

Yes type of this but better UI than this

[dry meadow]

Yes download button for both ios and Android will be there

• it will help users to get to know the app we can have a manual as to how the app works

With a contact us page

If @stoic igloo agrees with it we can start working on this

[brittle yoke]

It's a nice idea.

[loud skiff]

Idea looks great would love to collab

[warm bay]

I think we can go ahead with this

@dry meadow Great insight

[brittle yoke]

Check dm

[worthy citrus]

@brittle yoke You are now level 4! :confetti_ball:

[dry meadow]

https://debate-ai-prototype.vercel.app/

this is my idea for the landing page, Please have a Look

the last section has the download option

will not have login signup although but faqs and feature section can be there

will subtle animation and effects

[brittle yoke]

Do you think that is how it should be?🤔

[dry meadow]

Didn't get you?

[slow trail]

🚀 Hey everyone! I just opened an issue regarding input validation and error handling improvements in the Appointment API. I've also implemented a solution that enhances validation using Cerberus and improves error handling for better reliability.

🔗 Issue: https://github.com/AOSSIE-Org/BabyNest/issues/58
🔗 PR: https://github.com/AOSSIE-Org/BabyNest/pull/60

Let me know if you have any feedback! 😊

[dry meadow]

@stoic igloo waiting for your headsup

[stoic igloo]

Hey everyone,
Testing of the app is very crucial too

[stoic igloo]

There r some conflicts in the pr

[slow trail]

okayy on it

[crystal crag]

Greeting @stoic igloo , @proper musk ! I am Vanshika, a 3rd yr CSE student and new Gsoc contributor, as I worked with an open source community earlier, I liked the Baby nest project which is mentioned in Gsoc. With having a good command in React Native, Node.js, Python and SQLlite, I myself am capable of contributing in this project, can you please help me in doing so.

[rigid wedge]

Hey @proper musk and @stoic igloo ! 👋

I came across the BabyNest project and found it really exciting! 🚀 I'm a 3rd-year Computer Science student with experience in React Native, Node.js, FastAPI, and AI (LangChain, Python), and I’d love to contribute to this project.

I’m particularly interested in AI-powered milestone tracking, frontend development, and backend API optimization. Would love to know how I can get started and contribute effectively! Looking forward to your guidance. 😊

Thanks in advance! 🙌

Shah Dhruv

[terse elbow]

Heyyyy @dry meadow
Back from the exams 🙃
Could u give just a brief tldr

[dry meadow]

Hi @terse elbow that's great

[dry meadow]

We have worked on the model, tried to find the best one with minimum size and best accuracy

The model is working absolutely fine with offline access on local

[terse elbow]

Greattttt

[slow tree]

hello @stoic igloo and @proper musk I love the idea BabyNest and I have experience with React Native, FastAPI, and AI/NLP using LangChain, and I'd love to contribute in any way I can. Lookng Forward for your response

[slow burrow]

Hi everyone, I am a MSc. Artificial Intelligence student and along with that I have 5 years of software development experience in Flutter. Beside software development, I have also taught NLP with Deep Learning course to undergraduate classes for 1 year. My thesis in MSc includes fine-tuning and evaluation of LLMs for domain specific tasks. I also have done multiple Machine Learning projects using Pytorch.

[tribal nexus]

Hey @stoic igloo and @proper musk ,

I really loved the idea of BabyNest. And i think i can make significant contribution to this project because of my experience in React Native, Node.js. I am currently on my track of learning AI so i will surely learn more on the go and contribute better. I would love to be on board. Looking forward for your mentorship!

[stoic igloo]

Welcome everyone

[narrow night]

@stoic igloo I’ve sent you the draft for the enhanced features. Could you please take a look and share your feedback? I’d appreciate any suggestions or improvements you might have.

[narrow night]

@stoic igloo Sir, could you please check you dm once.

[charred patrol]

Hey @Bhavik and @Bruno 👋
I’m a B.Tech student in Electrical Engineering, and I recently came across the BabyNest project — I’m genuinely inspired by the positive impact it can have, especially in maternal health and community outreach.

I'm well-versed in technologies like Python, React Native, Node.js, and SQLite, and I’m always excited to learn more along the way. As a student, the journey of learning, contributing meaningfully, and seeing ideas turn into real-world solutions deeply motivates me.

I’d love to be a part of this initiative and contribute towards making BabyNest a success. Looking forward to your guidance and mentorship!

[torn tartan]

Hey @stoic igloo and everyone. I'm a first Btech computer science student specialising is Data science. This is my first time working with a open source project. I have good knowledge in languages like python, SQL, C, C++ and I am also currently learning java. I am a part of a social entrepreneurship club so just like that I want to work in fields where technology brings in a good social impact. I that way Baby Nest seemed really intresting to me. I have submitted my proposal and regardless of getting chosen or not I would still like to contribute. I am looking forward for guidance to help me contribute to this onderful community and learn various things .

[stone isle]

Hey @stoic igloo and everyone! 👋
I’m a B.Tech 2nd-year student and super excited to be part of this amazing community. I’ve already submitted my GSoC proposal and can’t wait to contribute and learn from all of you.

I’m pretty comfortable with Android development – I’ve built several projects using Kotlin and Flutter. I’m also very passionate about Deep Learning, AI, and Machine Learning, and I’ve participated in 20+ hackathons where I worked on various ML-based solutions. One of the projects I’m proud of is an app called SheSafe, designed to enhance women’s safety using smart features powered by AI.

Apart from coding, I love working on real-world problems and trying to bring tech-based solutions that can make a difference. I’m always curious to explore new frameworks, tools, and research areas – especially anything related to computer vision or intelligent systems.

I’m really looking forward to growing through this experience, getting valuable feedback, and building some meaningful contributions. I’d love to connect with others working in similar areas too — feel free to reach out if you’re into ML, Android dev, or just want to collaborate on something cool!

Thanks again for the opportunity — excited for the journey ahead!

[narrow night]

Hey @stoic igloo Sir
I’ve pushed the frontend part of the Medicine Tracker feature. Could you please take a look at the PR and let me know if any changes or improvements are needed?
Would really appreciate your feedback. 😊https://github.com/AOSSIE-Org/BabyNest/pull/67

[worthy citrus]

@narrow night You are now level 5! :confetti_ball:

[stoic igloo]

Will review it

[worthy citrus]

@stoic igloo You are now level 30! :confetti_ball:

[narrow night]

Okay sir

[fathom gust]

Hey AOSSIE community, Ankit this side. May I know more about the current status of BabyNest and what are the critical things mentors wants to get built in urgent? Looking forward to contribute...

[stoic igloo]

checkout issues

[fathom gust]

Okay sir

[charred patrol]

Hey, did anyone receive any mail regarding their proposal being accepted or rejected?

[warm bay]

Yeah, I have received acceptance mail

[stoic igloo]

Congrats @warm bay ✨

[dry meadow]

Congrats

[warm bay]

Hey everyone,
I just had a meeting with @stoic igloo . We discussed that our initial focus will be on developing the UI and implementing the agentic AI component to ensure the app functions effectively based on single-contextual data, along with an integrated notification system. The goal is to have this version ready for deployment by the mid-evaluation phase.

Following this, we plan to release subsequent versions of the app that will include additional features such as tracking capabilities, SOS functionality, and more.

So as discussed, currently I will be moving forward with the research part on how to implement the agentic RAG locally and efficiently. Once the approach is finalized, we’ll move ahead with development.
Will keep you all posted on the progress. Thanks!

[dry meadow]

Great!

[stoic igloo]

Yeh lets implement best context management for our local llm AI agent

[warm bay]

Sure

[verbal mist]

Just a doubt.! Are you guys are GSOC 2025 contributors or just an independent one.?

[warm bay]

Yes, @verbal mist I am a GSOC 2025 contributor

We are building a local context-aware assistant that:

1. Tracks and uses structured data (due date, location, health logs, appointments) via SQLite.
2. Retrieves relevant unstructured medical knowledge using ChromaDB as a local vector database.
3. Generates context-rich prompts dynamically for a local LLM (e.g., TinyLlama, Qwen, Mistral) to answer queries or make predictions.
4. Runs fully offline to ensure user privacy and support rural/inaccessible areas.

How we can do this:

1. Structured context - stored in SQLite example: location, due_date, appointments data
2. Unstructured context - stored in chromaDB example: vectorized pregnancy guidelines, appointment recommendations etc - filtered using metadata : week, location etc

Context management strategy:

1. Static structured - fetched from SQLite - includes profile details like due_date, location
2. Temporal structured - fetched from SQLite - includes appointments data (past and current) , weekly logs like weight (3 weeks rolling) etc
3. Unstructured - Vector embeddings + metadata - semantic retrieval from ChromaDB - Guidelines, vaccine schedules, regional policies etc - will give a domain knowledge to the llm example "Week 24 requires a glucose test in india"

Implementation:

1. LLM inference - llama.rn
2. Local structured database - react-native-sqlite-storage
3. Vector DB for retrieval - ChromaDB
4. State management - Zustand or redux

This is the basic plan of how we are proceeding with the agentic rag part. Any suggestions or feedback would be appreciated. Thanks

@stoic igloo @proper musk Should I go ahead with this?

[dry meadow]

Looks good 👍

[stoic igloo]

This looks good
Did you look up on storing llm contexts, so that we dont have to pass it everytime?

[warm bay]

Yes we can use context injection techniques i.e. can retrieve structured rows, top k unstructured items from ChromaDB after classifying the intent and routing

[warm bay]

Hi everyone 👋,

I’ve opened PR #69 – "Due Date Predictor Added" for review.
✅ What's included:

• New Parameters: We’re now collecting additional profile data like weight, age, and cycle length to improve due date predictions.
• Profile API: Added a get-set route to streamline profile data access and avoid redundant fetches.
• UI Update: Minor improvements made to the Home Screen for a cleaner user experience feel free to review and share feedback.

🛠️ Upcoming Work:

I'm preparing a follow-up PR for health tracking features (weight, mood, medications, symptoms — logged week-wise).
This will feed into our upcoming agentic RAG system, which needs:

• Structured profile data (from the new parameters)
• User health logs (from trackers)
• Verified medical guidelines (e.g., vaccine schedules, trimester tasks)

Once we integrate these into our local vector and relational stores (ChromaDB + SQLite), the agent will provide more personalized and context-aware insights — fully offline.

Please take a look at the PR and let me know your thoughts. Thanks!

[dry meadow]

LGTM!

[warm bay]

Hey everyone! 👋
Here’s a quick update on what we’ve implemented so far in the BabyNest AI agent:

• RAG (Retrieval-Augmented Generation) agent is up and running offline, combining structured data (SQLite) and unstructured context (ChromaDB embeddings).
• Supports personalized pregnancy milestone tracking based on location & due date.
• Handles appointments, vaccines, tests, symptoms, medicines, and user notes with week-wise tracking.
• Fully offline capable with local LLM integration for natural language queries & intelligent recommendations.

We want your input on a few important points:

• Additional Context Storage:
  What other pregnancy-related context should we store for better AI recommendations? (e.g., nutrition info, exercise logs, emotional well-being, partner involvement, etc.)
• Orchestration Approach:
  Should we build our own custom orchestration logic for the agent workflows, or adopt a framework like LangGraph for modular, graph-based orchestration? What pros and cons do you see?
• Other Key Considerations:
  Any suggestions on improving offline performance, user experience, or expanding the agent’s capabilities?

Feel free to reply here or DM me directly!

Special request to our admins and maintainers:
We’d really appreciate your insights on the overall architecture and roadmap to ensure we’re aligned with BabyNest’s vision and priorities. Your guidance will help us focus efforts where they matter most. @stoic igloo @proper musk

[dry meadow]

I think working with conventional frameworks like langgraph is better

[lyric berry]

adapting graph based framework would be better ig (LangGraph)
kind-of ensures ease of scalability to the system and code quality too will be a lot modular and organized.

[worthy citrus]

@lyric berry You are now level 14! :confetti_ball:

[stoic igloo]

Why do we need orchestration
Our model is very simple
Like setting reminders for tasks that needs to be done
Use a simple custom logic for that

Include a video demonstration of what’s been done so far

We should also become platform independent
Making sure our app works in both ios and android

And Lets make a roadmap to launch the play-store version first
Give me a list of pending tasks and tests before we can launch the v1

[warm bay]

Ok then ig we can go forward with langGraph for the agent

[warm bay]

I can raise a pr for what has been done yet along with the video

[warm bay]

Yeah that we have to see - to test it on iOS

[stoic igloo]

upto you, you shouldn't need langraph for a simple model

[hushed ruin]

See what i just built 😍
https://x.com/nest_promp15702/status/1947280078016070117

PromptNest (@nest_promp15702)

Built this AI system to save 15 minutes every day. Took 5 hours with @make_hq . No regrets.

Screenshot👇
#Automation #AI #tool #Innovation #Growth

X · Jul 21, 2025, 12:58 PM

[azure hemlock]

Found X link(s)! Here's the mirror using xcancel.com:
https://xcancel.com/nest_promp15702/status/1947280078016070117

[hushed ruin]

what is this?? xcancel??

Interact Guys 😍
https://x.com/nest_promp15702/status/1948569487139106844

PromptNest (@nest_promp15702)

I went silent for 3 days. Here’s what actually happened behind the scenes — and why it made PromptNest stronger.

X · Jul 25, 2025, 2:22 AM

[cunning wedge]

I'm Md Irshad, skilled in HTML, CSS, JavaScript, ReactJS, Flask, and SQL. I'm excited to contribute to this project and collaborate with you all.
If there's anything I can help with—especially on the frontend or backend—I'd love to get started!

[warm bay]

Hey @cunning wedge cheers for joining the community

You can look for some better designs and share it with us, will be really helpful

[hushed ruin]

https://www.instagram.com/p/DMqFUURyS_l/?igsh=bjVxbTg1MThleGxu
Interact guys 🥰

promptnest.brand

Gemini 2.5 Pro isn’t just another AI tool — it’s your multi-talented assistant 🤝

I’m breaking down complex AI tools into simple language for creators and solopreneurs.

📩 DM “Gemini” if you want my free beginner guide.

#GeminiPro #AIExplained #PromptNest #SundayLearning #AITools

Instagram

[hushed ruin]

https://www.instagram.com/p/DMsqGROyAvS/?igsh=bzJmZWo4YWlvYnln
Interact guys 🥰

promptnest.brand

🚨 Want to build your own AI agent — no coding needed?

Here are 5 tools you can start with TODAY to automate:
✅ Client onboarding
✅ Customer support
✅ Social media tasks
✅ Lead generation

Tools Covered:
1️⃣ Relevance AI
2️⃣ Make.com
3️⃣ Vapi.ai
4️⃣ Agent.so
5️⃣ Dify.ai

💡 Want me to break each one dow...

Instagram

[warm bay]

Hi everyone,
I have raised a pr which addresses the issue where the BabyNest agent recomputed context and regenerated embeddings on every query, even when the underlying SQLite database had not changed. This led to unnecessary computation, disk I/O, and degraded performance — especially in offline or resource-constrained environments.

I have implemented a database hash based staleness detection method and now the cached context in stored in memory for reuse purposes. I have also implemented a auto-trigger vector store refresh functionality. Also added a test file for the agent.

https://github.com/AOSSIE-Org/BabyNest/pull/78

Now I’m at the point where I want to integrate this agent into the React Native frontend, so it can drive the entire app — interpreting user queries, triggering UI changes, and responding naturally.
Any suggestions or examples for:

• Structuring agent → UI actions (e.g., navigation, showing modals)
• Using a single /agent/process route instead of multiple endpoints
• Efficiently passing structured + semantic context into the local LLM

Would really appreciate any insights, patterns, or lessons learned!

[proper musk]

@warm bay , I saw that @stoic igloo reviewed your PR 2 weeks ago. Did you get everything that you needed from the review?

[warm bay]

Yeah @proper musk, I got everything I needed from the review and discussed the same with @stoic igloo too

[stoic igloo]

@warm bay

Hi vivek
I went through the changes you committed
Why are we invalidating cache when there is an update in db?
We should update the cache instead whenever there is a change in db
Create an update_cache method for it which captures the change and update accordingly

Make sure to add a check so that this file doesnt go over 70% size limit for context and we update and delete unnecessary cache accordingly

Other changes look fine
Include screenshots/demos for the new screens and their working (recommended actions screen)

[warm bay]

Hey @stoic igloo
What we are doing right now is when there is any change in db, we are invalidating the cache and on next query we are rebuilding it which ensures that cache remains fresh after updates

[worthy citrus]

@warm bay You are now level 23! :confetti_ball:

[warm bay]

We can also follow the update_cache approach

I'll do the changes if we are planning to follow this

[warm bay]

Have done the required changes @stoic igloo

Please review

[stoic igloo]

i can only see the tests folder in your commit
I cant see update_cache defined anywhere, there must be something wrong with your commits

[stoic igloo]

yeh, please fix it
as invalidating the cache and rebuilding it completely again everytime doesnt make sense

[warm bay]

Let me check

[warm bay]

Hey @stoic igloo , the update_cache implementation is already present in ContextCache class in cache.py

Have done another commit, please check now

[stoic igloo]

Oh great, I will review it soon

[stoic igloo]

@warm bay
I saw the PR
Most of it looks good
I will merge it so we can move ahead
Can you also include how much response time has improved after implementing new cache methodology? Include a video for the same, interacting with the app
Also how is this method working? Current_cache.update(profile_date)
few concerns that you can improve in next PR -
Improve code quality
Clean the code and Keep it very simple
I saw 4 functions with clean cache functionality
Improve overall readability of code, remove redundant functions and remove all print statements too
We can start focusing on building the frontend completely now, and integrating it with the timeline data and others
Lets do this, and we can release a beta apk for testing

Improve file structure too
Lets make subfolders to handle particular tasks like cache management or agent response

[warm bay]

Yeah sure

[warm bay]

Yeah starting with the frontend part

[stoic igloo]

Any updates?

[warm bay]

Working on it

Busy doing debugging on the context passing part

[stoic igloo]

Any updates?

[true creek]

Hi everyone! 👋
I’m Nived, a first-year electronics and telecommunication engineering student. I’m really interested in learning more about programming, open source, and building practical projects. I’m still exploring and improving my skills, but I’m excited to collaborate, learn from the community, and contribute wherever I can. Looking forward to connecting with you all and growing together!

[warm bay]

Actually I am still caught up in the debugging part. Have done the chat automation with agent, currently debugging the tasks integration with the agent - there are some issues I am facing,but will do it asap

[stoic igloo]

@warm bay I have added comments

[bitter shadow]

Heyy @stoic igloo I'm really interested and I'd like to contribute to this project, I can contribute in the frontend part i know react

[stoic igloo]

That’s great
Try setting up the project

[bitter shadow]

@stoic igloo I'm having some issues while setting up the project, can i d m?

[stoic igloo]

Pls share ur issues here

[bitter shadow]

it says no module named chromadb found but i already installed it

in the backend folder

no issues while installing dependencies in the frontend

[warm bay]

Try installing it in virtual env

[pallid nacelle]

Hi everyone, I'm new to this project. Could someone please help me set up the codebase? Also, I'm a full-stack developer, so I'm looking for good first issues related to web development—frontend, backend, or full-stack. If the project is primarily about mobile or machine learning, it might not be the best fit for me. Thanks!

[river arch]

@stoic igloo hello is this project currently active ???

[stoic igloo]

Project is active
But final list of selected projects will be released later by next year

[river arch]

@stoic igloo i have raised a issue kindly assign it to me i have stared working already

and do we have to wait for the maintainer to assign as the issue or we can start makeing pr direactly without getting the issue assigned from the maintainer

???

after raising the pr for this issue i will started working on the issue the use have created regarding the model

[worthy citrus]

@river arch You are now level 2! :confetti_ball:

[upbeat moon]

@stoic igloo hello everyone.. This side Suryansh Garg... From IIT BHU.. I am new to this project... I find it very interesting... Can you guide me through like... How should I start contributing to this project?

[warm bay]

Hey @upbeat moon, good to know that you are interested. Firstly please try to setup the app using the guide then you can move on to find issues

[upbeat moon]

Suree..

[river arch]

@warm bay @stoic igloo
i have raised a pr please review it

[river arch]

@proper musk i have raised a pr please review it

[robust aspen]

Hi. New contributor here. I know this question irritates a lot.

But can someone please tell me, willl this project be there in GSOC 2026??

[eternal tapir]

Hi 👋
I’m new to open-source and really interested in contributing to Babynest. I’ve gone through the repository and would love to get involved could you please suggest some good first issues or small enhancements that would be a good starting point?

Also, I wanted to ask if Babynest is planning to participate in Google Summer of Code 2026 (GSoC 2026), or if there’s any plan for it. If yes, I’d love to know about the next set of features or areas you’re focusing on so I can start learning and aligning my contributions in that direction. @stoic igloo @proper musk

[stoic igloo]

Yep
Shortly

[stoic igloo]

@river arch added a comment

[worthy citrus]

@stoic igloo You are now level 34! :confetti_ball:

[strange wasp]

Hey everyone! 👋
I was going through AOSSIE’s list of projects and came across BabyNest. I really love what you’re doing here! There are so few apps that focus on supporting women through the complexities of womanhood (the hormones part alone deserves its own textbook 😄).
I’ve started setting up the project locally to explore how it works, and I’m really excited to contribute and be part of this community!

[devout tusk]

@warm bay @stoic igloo @proper musk I noticed the last update on the Babynet repo was about a month ago.
I’ve been exploring the project for a while and wanted to check whether the current focus of the org is still on this repo or has shifted to another more important project. Would love a bit of clarity so I can contribute effectively.

[robust wave]

Please sir guide me gsoc program

[warm bay]

Hey @devout tusk, yes this project is in org's focus rn, you should try and setup the project and start contributing. Cheers

[warm bay]

@strange wasp, it's good to know you find the project helpful and enthusiastic about contributing to the repo. Try to setup the app and share any challenges you face on this grp

[strange wasp]

Hey! I’ve been exploring the app after setting it up (I ran into a small setup issue but got it working in the end). The app looks clean and minimal, which I really appreciate.

I had a small question: are UI/UX suggestions welcome at this stage?

I noticed that the app currently uses a lot of pink for its highlight/accent color. I was wondering if the team is open to discussing alternative accent colors. For example, a softer red or pastel red might give a more balanced contrast on a white background while still keeping the warm, welcoming tone of the app.

This is just an idea, and I’d love to hear the maintainers’ thoughts before creating an issue or suggesting anything more formally.

[tropic fiber]

hello respected sir\mam i am here to contribute to your project how can i help you

[stoic igloo]

Definitely we need to improve ui ux
Especially making sure that they are compatible with both ios and android

[strange wasp]

So I can raise an issue regarding this? With the suggested colours and changes in the design?

[lethal socket]

Hi @stoic igloo
I have find a UI related issue
regarding Sidebar.
I would like to work on this.

[storm heath]

Hey team! 👋 .I came across BabyNest while exploring AOSSIE’s projects and instantly loved the vision behind it(mainly unclustered point to point services) . I’ve started setting up the project locally and would love to contribute. I have experience with React Native and Node.js and a keen interest in this project

[lethal socket]

Hi
I have some doubts regarding Sidebar and bottom navbar
Why we need both of them ?
Bottom Navbar is sufficient.

[strange wasp]

Sidebar has links for extra details and bottom navbar is for more basic navigation from what I could understand. I feel like the side navbar should cover the entire screen when opened. This helps avoid the clash of both side and bottom navbar. This also helps to prevent the confusion of two navbars being displayed at the same time, though only a small part of the bottom navbar is visible when the side one is opened.

[proper bolt]

HI all! I am just diving into the contributions here , may i Know what work or help required from my side and which domain of project are your guys currently working in ! Your guidance is really appreciated since its my first time here !

[lethal socket]

Why there is two chatbot buttons ?

[strange wasp]

They most likely do the same thing. We could check it once to confirm.
If they end up being redundant, maybe one of them can be removed, of course, if the maintainers think that makes sense.

[lethal socket]

One last Question.
How we are saving the user data?
A user couldn't remembers the date for all time .
What happen app get crash ?
We need email or phone to map user data.

Why we are deleting user data on logout ?

[strange wasp]

From what I’ve understood so far while exploring the backend architecture, this app is not really authentication-based in the usual sense (like email/password or phone login).

It works more like a local, data-centric app. When a user opens the app, they enter basic pregnancy-related details such as last period date, cycle length, etc. Based on this information, the app calculates things like the current pregnancy week, due date, and shows relevant guidance, appointments, and logs.

All this data is stored locally in a SQLite database. There isn’t a concept of permanent user accounts. If the user logs out or resets the app, the stored user data is cleared.

So the app is mainly centered around the current user’s data and pregnancy timeline, rather than user authentication or identity management.

[warm bay]

@strange wasp Yess exactly

This is more like a personal app no auth no cloud required

[wintry barn]

Hi everyone 👋
I’m Vatsal, a 3rd-year bachelor’s student , Joining Swiggy as an AI/ML Intern and an active open-source contributor.

I’ve been going through projects under the AOSSIE group, and InPactAI, PictoPy, and BabyNest really stood out to me in terms of impact and alignment with my interests in AI/ML and applied systems. I’m planning to prepare early for GSoC 2026 and would love to start contributing meaningfully ahead of time.

If there are any maintainers or contributors here, I’d really appreciate guidance on:

• beginner/intermediate level issues or areas to start with
• recommended issues and features i can contribute to

[feral compass]

Hi everyone
I'm Alizah, sophomore from Hyderabad, Ind
I’m currently exploring AOSSIE as a potential organization for GSoC 2026.
I’m interested in contributing to beginner-friendly projects and getting familiar with the AOSSIE workflow, codebase, and community. Right now, I’m starting with documentation and small feature/bug-fix contributions to learn and improve step by step.

lets connect!!
https://www.linkedin.com/in/SyedaAlizah/

[grim badger]

Hi @stoic igloo,
I’m new to open source and after setting up the repo locally, I tried to fix a bug with the drawer “Tasks & AI Recommendations” button — it now correctly navigates to the AllTasks screen.
The PR is open here: https://github.com/AOSSIE-Org/BabyNest/pull/123

[lethal socket]

Any Suggestion

it`s old one.

Hi @stoic igloo
Please comment on it.

[worthy citrus]

@lethal socket You are now level 2! :confetti_ball:

[lethal socket]

Thanks .

[lethal socket]

Both are same.
Can we drop one or both are important. ?

Hi @stoic igloo
I have doubt regarding it.

[stoic igloo]

We can just keep this weeks tasks on home page which also leads to full info abt tasks

And remove 3 columns above it

And calendar can be separate

[warm bay]

Looks good @lethal socket

[lethal socket]

Can we integrate calendar and timeline in one page ?

Hi @stoic igloo
@warm bay
Please give your thought on it.

[warm bay]

No @lethal socket we can't do that. Calendar is for appointments and timeline is for tasks

[lethal socket]

Ok 👍

[lethal socket]

Hi @stoic igloo
Please review it and give your feedback
if you some want change then I will do it.
https://github.com/AOSSIE-Org/BabyNest/pull/127

[worthy citrus]

@lethal socket You are now level 3! :confetti_ball:

[lethal socket]

Recently I explore pregnancy+ app.
It UI is simple and user friendly.

[lethal socket]

Merry Christmas 🎄 Guys.

[storm heath]

Merry christmas❄️🎁 Guys

[storm heath]

Great , it look simpler and user friendly to me also

[lethal socket]

@stoic igloo @proper musk @warm bay
https://github.com/AOSSIE-Org/BabyNest/issues/133
I have proposed a feature .
please read it and give your suggestion.

[scenic urchin]

@stoic igloo We can add a feature to automatically fill the user's country in the login page.

Also, we at present user has to scroll through countries to find her. We can implement search there, if user types some first letter of her country we will filter those

I have created a issue for this
https://github.com/AOSSIE-Org/BabyNest/issues/138

[stoic igloo]

Yep go ahead
Implement auto fill textbox

[scenic urchin]

can you assign me this so that i could start working

[scenic urchin]

@stoic igloo I want to discuss about the networking issue i am facing. I have created issue 137 for it

My platform
os - macOS
platform - android studio

Issues

1. The frontend is sending request to backend on BASE_URL=http://192.168.1.4:5000, which doesnot work unless i run adb reverse tcp:5000 tcp:5000 shifts localhost from the virtual device to my system

Solution - Mention in setup.md to run adb reverse command in the backend section.

2. React-native caches .env file and networking settings due to which even after running adb reverse it doesnot work

Solution - need to clear cache before starting app again using npx react-native start --reset-cache. Need to mention this in setup.md backend issues

3. The file Frontend/src/services/ConversationContext.js uses wrong way of fetching environment variables, const BASE_URL = process.env.BASE_URL || 'http://127.0.0.1:5000'; doesnot in react-native.

solution - We need to do import { BASE_URL } from '@env'; at the top of the file

I have done PR for this issue, implementing all these solutions

[strange wasp]

I faced the same issues. The base url needs to be changed according to your ip address

[worthy citrus]

@strange wasp You are now level 2! :confetti_ball:

[strange wasp]

The one already present in the env file is hardcoded

[scenic urchin]

Yup, i have done PR 145 that solves this, have a look

[worthy citrus]

@scenic urchin You are now level 2! :confetti_ball:

[scenic urchin]

Implemented this feature in PR 148

[stoic igloo]

Ohk I will take a look

[hardy bolt]

Hi everyone,
I’m currently exploring the BabyNest project under AOSSIE and going through its repository and documentation.
I’m interested in starting my first contribution here.
Could you please let me know if there are any beginner-friendly / good first issues or small features or documentation improvements that I can start with?

[strange wasp]

you’ll have to check out the issues section on GitHub

[scenic urchin]

@stoic igloo I request you if you could guide me here. I will happy if you could be maintainer here as well

#general message

[strange wasp]

@stoic igloo I just wanted to get a bit of clarification here. So, I noticed that in the backend folder, we have a database.db file that basically stores all the SQL data. That makes sense. I also looked at the schema.sql file, and I can see that it’s creating the tables and inserting mock data. But none of this data actually has a user ID associated with it.

It clearly looks like the appointments and other things aren’t meant to be multi-user, otherwise they would have a user ID.

Here’s where I’m confused: when we call different methods in the code, there is a user ID being passed, like in the update_cache method. So I’m not sure, is this app meant to store all user data in a separate file on each user’s device, so that each one only holds their own data? Or is it meant to store all user data in the backend database.db file?

If it’s the latter (multi-user with a shared database), then all the appointments and other tables should have a user ID, otherwise it doesn’t make sense. But if the data is meant to be stored separately on each user’s device, then I’m a bit lost because of the way update_cache works.

Basically, I’m really confused about this. If the user ID is not meant to be used, then why is it there? And if it is supposed to be used, then it should definitely be in all the tables like weight, medicine, symptoms, blood pressure, etc, but it’s not. According to the SQL, it looks like there’s only one user, and all the tables are just for that one user.

Just wanted to check this out so I know how to proceed with my assigned issue.

[stoic igloo]

Hey this concern is valid
We shouldn’t need user id
As there can only be one user

But if it’s not breaking anything
It shouldn’t matter

Data is stored locally on users device

[strange wasp]

Thanks for the clarification. I noticed there’s a single database.db file in the backend, is this intended to store data for all users, or is it mainly for local development/testing?
Just wanted to understand the design choice here.

[hardy bolt]

@stoic igloo
I was setting up the backend locally and ran into a ModuleNotFoundError: No module named 'chromadb'.
After installing chromadb manually, my Flask server started successfully at http://127.0.0.1:5000.
I noticed there’s already an open GitHub issue about backend startup problems, missing dependencies, and setup docs, opened three weeks ago, but it’s unassigned.
It seems this problem is still reproducible.
I wanted to ask:
Is chromadb a required dependency that should be added to requirements.txt?
Would it be okay if I work on fixing this and updating the setup docs if needed?

[stoic igloo]

I think it was resolved

[stoic igloo]

Only one user data shud be stored
It might be for testing

[strange wasp]

Oh okay. Thanks so much for the help.

[uneven linden]

hi @stoic igloo
actually when the backend agent is turned off, here the frontend didnt really know what to do. As i made, when the situaltions are this, the backend agent exposes a simple /health check. the UI shows a small warning banner instead of failing silently. normal chat falls back to local responses, but structured features (appointments, health tracking, etc.) still work as usual. then the app stays usable both in dev and prod even when the agent isn’t available. I have opened a PR #134 around handling this. Please review and acknowledge for improvements if any...

[lethal socket]

some UI changes.
Guys please give your feedbacks.

[strange wasp]

looks simple and minimal

[scenic urchin]

If you could give a border for bot message as well, it will be best

Looks good

[main tiger]

Hi there 👋,
I’m Lokesh Somaiya, a B.Tech Electronics & Computer Science student from VESIT, Mumbai. I’ve been actively working across full-stack development (MERN), AI systems, and backend engineering, along with hands-on cybersecurity industry experience.
I am highly interested in contributing to AOSSIE’s open-source projects and I’m also preparing seriously for Google Summer of Code 2026. I’d love to get involved, understand the project expectations, contribute meaningfully, and learn from the community.

It would be really helpful if you could guide me on:
• How I can begin contributing effectively to AOSSIE
• Suggested repositories or project areas suitable for my tech stack
• Best practices and expectations for first-time contributors

I’m genuinely excited to collaborate, learn, and contribute to impactful work. Your guidance would mean a lot 🙏

[scenic urchin]

I noticed a bug while exiting the navbar window without using the cross button

I have created an issue for it and i am working to resolve it.

https://github.com/AOSSIE-Org/BabyNest/issues/153

[scenic urchin]

Hey @stoic igloo, I am working on this from last two weeks and i have created these issues and PRs please review and merge them if possible

[worthy citrus]

@scenic urchin You are now level 6! :confetti_ball:

[livid drift]

Hi @stoic igloo , I have identified an architectural gap in the AI Agent logic where handlers for Medication and Blood Pressure are missing, despite having existing backend routes and database tables. I've documented the inconsistency and the impact on AI chat in this issue: https://github.com/AOSSIE-Org/BabyNest/issues/156
I'm currently working on the implementation to bridge this gap.

[livid drift]

Hi @stoic igloo Mangla, I've just updated the PR #157 for issue #156.

I have refactored the logic to include entity extraction (regex) for medicine names and BP readings, fixed the intent classification in intent.py, and resolved potential database connection leaks. The code is now fully verified and bot-cleared.

Looking forward to your review! Link: https://github.com/AOSSIE-Org/BabyNest/pull/157

[strange wasp]

@stoic igloo the architectural diagram of BabyNest mentions the use of react-native-sqlite-storage, but I couldn’t find its implementation anywhere. Is it left to be completed?

[livid drift]

Hi @stoic igloo , I have successfully implemented the AI Agent handlers for Medication and Blood Pressure logging.

I have addressed all the feedback from CodeRabbit, including:

Security: Implementing dynamic and persistent User IDs for privacy.

Stability: Adding robust URL validation and backend retry logic.

UX: Fixing Android keyboard avoidance and accessibility issues.

All CI checks are now passing, and the PR is ready for a final review: https://github.com/AOSSIE-Org/BabyNest/pull/157. Thanks!

[worthy citrus]

@livid drift You are now level 2! :confetti_ball:

[scenic urchin]

@stoic igloo I have uploaded videos and images of changes i have as requested by you in the respective PRs

Also, why my PR which was done before Naren was rejected and his was accepted even it was not associated with an issue.

https://github.com/AOSSIE-Org/BabyNest/pull/148

[scenic urchin]

I have resolved this issue in two PRs PR 167 (profile icon) and 168 (notification system)

https://github.com/AOSSIE-Org/DebateAI/issues/155

[stoic igloo]

Reviewed many PRs
Pls ping me if something is updated or pending

[lethal socket]

@stoic igloo
Please see the recent pull request regarding the Dependency change.

[worthy citrus]

@lethal socket You are now level 4! :confetti_ball:

[lethal socket]

How many are working on ChatScreen ?

Let work Together.

[livid drift]

Hi @stoic igloo , I've submitted a fresh and clean PR for the medicine and BP logging handlers. I have completely reset the branch history to remove all accidental junk files from the previous attempt. It's now ready for review with only the 6 core file changes.
https://github.com/AOSSIE-Org/BabyNest/pull/164

[strange wasp]

@stoic igloo I wanted to get some feedback on issue #161 regarding fine-tuning the Qwen-0.5B model. The issue hasn’t been assigned to me yet, but I’ve started preparing a small sample dataset to better understand the requirements and direction.

My current plan is to keep the dataset to around 150 samples for initial fine-tuning of the SLM. I’ve drafted a preliminary version of the dataset and would really appreciate any feedback on whether the format, size, or approach should be adjusted before I proceed with training.

Please let me know if there’s anything I should change or improve. I’m happy to modify it accordingly. Thanks!

This is the dataset as of yet

[lethal socket]

@stoic igloo
I'm cleaning up all those scattered fetch calls and moving them into a dedicated src/api/ folder. This will make the code way cleaner and easier to manage for everyone.

src/
└── api/
├── client.js --> The base fetch wrapper (interceptors, base URL)
├── appointments.js -->Appointment-specific endpoints
├── health.js --> Combined logic for weight, medicine, BP
├── user.js --> Profile and task-related endpoints
└── agent.js --> Agent-specific context and RAG endpoints

please see the folder structure and give your feedback .

[scenic urchin]

@stoic igloo I have added screenshots and videos where required in my PRs and edited them to match PR format. Can you please review them

PR 144, 145, 146, 147 and 154

[scenic urchin]

@lethal socket Have you just improved the UI of chatbot or improved its backend. Because at present its terrible at processing requests

[lethal socket]

I work only on frontend side.

[scenic urchin]

Ok then i will proceed with improving its backend

[lethal socket]

Ok

[lethal socket]

Anyone is interested to work on voice feature for ChatScreen.

Let work together.

[strange wasp]

are u talking about the backend part?

[lethal socket]

Frontend

[strange wasp]

If there is any backend part, I can work on it with you

[scenic urchin]

This might be overkill, we might not need a voice feature there

[worthy citrus]

@scenic urchin You are now level 11! :confetti_ball:

[livid drift]

Hey everyone! I was following the discussion about the ChatScreen features. I think adding OCR (Optical Character Recognition) would be a great addition.

[livid drift]

Parents could scan medical reports or prescriptions, and the AI (LangChain) could automatically log dates or symptoms.

LMK if anyone is interested in this or has any feedback?

[scenic urchin]

@livid drift This will be good feature but OCR is not enough for this task. We might need a ML model to extract useful info from OCR extracted text

[livid drift]

Exactly @scenic urchin ! My plan is to use Google ML Kit for the base OCR and then pass that extracted text to our existing LangChain/LLM backend. The LLM can easily perform Named Entity Recognition (NER) to structure the unstructured text into JSON (like extracting dates, doctor's notes, etc.). This way, we don't need to train a custom ML model from scratch. Thoughts?

[scenic urchin]

We donot have any LLM in our backend

Chatbot works on regex patterns

[livid drift]

Fair point @scenic urchin I noticed the current logic is Regex-based, which is great for speed. But for OCR extraction, Regex might be too rigid since medical reports vary so much.
How about we use a Hybrid approach? We can call a lightweight, open-source model (like via Hugging Face Inference API or Gemini free tier)

[scenic urchin]

Which OCR can understand a doctor's writing 😅

this feature is doomed

[livid drift]

I get the concern, handwriting is tricky. But that’s why I suggested a Hybrid Approach.

[worthy citrus]

@livid drift You are now level 3! :confetti_ball:

[scenic urchin]

this might work on online soft copy reports, but i don't think doctors give patient online prescriptions

[livid drift]

That's exactly why OCR is needed, If everyone had soft copies, we could just parse the PDF. The goal of using Google ML Kit is to bridge the gap between 'Handwritten Paper' and our 'Digital Database'. i know it might be difficult but this is a very common use case in Health-Tech apps today!

[azure root]

Hieveryone, i am saksham and ive been lurking here for past few days. I am a second your CS undergrad from IIIT Hyderabad. but i am very new to open source and am unable to get how can i start contributing like can anyone tell. it would be very helpful

[livid drift]

Welcome to the community, Saksham! 👋 I'm Pranjal. Don't worry, even I joined very recently.
I just finished setting up the project on my M4 Mac, and it can be a bit tricky due to some environment issues. I’d be happy to help you get the local setup running so you can start exploring the codebase. Feel free to reach out if you get stuck anywhere!

[strange wasp]

The regex is too rigid. I myself have been trying to fine tune the existing qwen-0.5b model for proper extraction of appointments. There is no proper dataset in the internet for fine-tuning and that limits the extensibility of the model. On the other hand, the model is loaded in the user’s device so even upgrading it is a problem because of size issues.

[scenic urchin]

The optimal way is that we improve the regex

I am on it, already have done it for making appointments

By evening, I will make a PR

[strange wasp]

I improved the regex quite a lot for the appointments too and the pr got merged but still the maintainer said that the ultimate way is to fine tune the slm. I guess if all of us work on making a large dataset for extraction of appointments, symptoms and the like, fine-tuning can really help

[scenic urchin]

Ok I will look into, let's work together 👍

[strange wasp]

Yeah sure

I’ll make the pr today containing details of the fine-tuning I have done till now. You can look into it

[livid drift]

That makes total sense @strange wasp . If the maintainers prefer an SLM-based approach, then building a robust dataset is definitely the priority.I’m fully in to help with this!

[strange wasp]

Yeah let’s build one together. Even if we have 1000 samples, that will really help

[worthy citrus]

@strange wasp You are now level 4! :confetti_ball:

[livid drift]

Hey @strange wasp , I've drafted the first batch of 20 samples covering appointments, symptoms, and some messy OCR-style medicine logs. I tried to keep the format consistent with yours. LMK if this looks good

[strange wasp]

Okay I’ll check it and let you know by tonight

Tho at this point I am only fine-tuning for appointments extraction to see how it works out. So you can leave the rest of the data samples for now

[scenic urchin]

@stoic igloo What could be max size of this model we are talking

Also, do we want the bot to have medical knowledge or just be able to schedule appointments and fetch data

If the bot would recommend user, like answers to questions like "Can i eat sushi?", this is very critical

I don't think anyone should rely on chatbots for things like this

[strange wasp]

I think that maybe this should already be mentioned somewhere that bot is not for informational purposes rather just for helping people keep up and informing them about any abnormality in their data

[stoic flame]

We can't transcribe doc's handwriting, but we can get medical history from medical bills, which are comp. Generated usually

Hence we can get medical history and log them, store medical prescription as image, and again, medical reports are usually computer generated so we can do that too.

It will be more like a medical history record keeping

Additionally we should also add a export feature (idk if it exists didn't check) so when they have to quit app, they can get clean copy of everything, usefull for serious illnesses/diagnosis

[strange wasp]

Yeah this is alright. But for any missing data for the date use XX’s and not make up missing data since when I did that the model started making dates up.

[alpine pawn]

Hi
​My name is Venkat, and I’m a student developer preparing for GSoC 2026. I'd love to contribute.
​I want to propose an original project: 'Proactive Vision' (Smart Glasses for the Blind). Unlike existing glasses that require buttons or voice commands, my software stack uses low-power edge AI (Tiny-YOLO) to automatically detect when a book is placed in front of the user. It specifically solves the 'distortion' issue in curved pages to ensure high OCR accuracy and uses Spatial Audio to anchor the reading voice to the book’s physical position.
​I believe this proactive pipeline would be a great addition to the Zendalona ecosystem. Does the team encourage student-proposed ideas like this for GSoC 2026?. I wish to contribute this idea for your organization

[strange wasp]

I think there is another channel for such discussions. I don’t think you will get much insight here. Though this idea sounds good.

[alpine pawn]

Can u say name of the channel

[scenic urchin]

@stoic igloo Please review my PRs

PR 144, 145, 146, 147 and 154

[strange wasp]

I haven’t gone thru all the channels. Maybe it’s the tech discussion one.

[livid drift]

Hey @stoic igloo , Just a quick reminder to please check the proposal for Issue #165 (Smart Medical Alerts & Anomaly Detection) whenever you get a chance.

[livid drift]

Hey @stoic igloo , I'm starting work on the OCR feature for appointment and prescription extraction.I’m planning to use Google ML Kit (Text Recognition) because it works entirely on-device,Do you think this is the right approach for the project.

[strange wasp]

@stoic igloo I have submitted a thoroughly documented PR 170 regarding the process of slm fine-tuning for qwen 0.5b. Please check it out when you get time.

[livid drift]

hi @strange wasp I've added edge cases like typos, ambiguous timings (e.g., 'checkup at 5'), and shorthand queries that real users actually type. This should help the SLM handle noise better.

[worthy citrus]

@livid drift You are now level 4! :confetti_ball:

[strange wasp]

We’ll add this data once the maintainer reviews the pr I have submitted regarding fine-tuning the slm

[silent pike]

Hi @Naren,
I’m @dheeraj12347 from GitHub, working on BabyNest Issue #151 (frontend refactor / 5-layer architecture).
I’ve already:

set up a centralized API client (src/services/apiClient.js), and

created hooks like useWeight and useAppointments in src/hooks/.

I’m now trying to refactor the existing UI to use these hooks, but I’m having trouble finding the components that currently handle:

baby weight / growth tracking, and

appointment creation / management.

Could you please tell me the exact file paths for the weight screen and the appointment UI (e.g., screens or modals) so I can wire the hooks into the right place without breaking anything?

[lethal socket]

Hi @stoic igloo
Need your guide .
For this.

[worthy citrus]

@lethal socket You are now level 5! :confetti_ball:

[lethal socket]

I'm little bit confused
BabyNest is an on device app.

Why we created a backend?

We can do query call sqlLite in frontend only.

If We need backend ,
How we will manage both in an apk?

[lethal socket]

@stoic igloo
@proper musk
@strange wasp
@warm bay
Could anyone help me ?
I want to know ...

[strange wasp]

From my understanding, the backend is currently being used primarily for testing and prototyping purposes. We are creating a replica of the intended frontend SQLite interactions, so all database operations are handled in the backend for now.

Additionally, the backend is responsible for creating separate user context files, which store user data both in memory and on disk. These context files help reduce unnecessary database calls.

The backend also handles tasks such as extracting structured data from user queries and managing agent-related logic.

[worthy citrus]

@strange wasp You are now level 5! :confetti_ball:

[lethal socket]

Thanks @strange wasp

[stoic igloo]

Hey guys, let’s make sure we don’t raise unnecessary PRs, AI slops and help each other out. (Every contribution can be included in proposals, not just the PRs) Try to understand the vision of the project, future prospects and also the codebase to contribute effectively.

I will try to review them once I find time. Have patience and good luck

[scenic urchin]

@stoic igloo Thankyou so much for your precious time Bhavik, i will go trough my PRs and remove not that important ones

[livid drift]

hi @strange wasp, I've been working on the OCR and Vitals Extraction feature for our project. I have successfully set up the pipeline using OpenCV for preprocessing and Pytesseract for text extraction.
However, I'm finding that the base Qwen-0.5B model is struggling to accurately parse noisy medical data, often returning null or incorrectly formatted results.
I know you've been fine-tuning an SLM for this project. Would you be open to sharing your fine-tuned model (or a GGUF/LoRA adapter) so I can integrate it into my current pipeline? I believe this synergy would significantly improve the accuracy of the feature.

[strange wasp]

Yeah sure. I have already opened a pr containing a complete description of my attempt to fine tune the model. I am waiting to hear from the maintainer on that one. You can check it out, it’s pr number 170

I have not created a GGUF file as of yet. I fine tuned the base model only. Once it gets accepted, then we can make it into a GGUF file. All the links to the LoRA adapter and datasets are provided in the pr only. If you find a problem, do inform me.

[livid drift]

Thanks, I'll let you know if I need something

[uneven linden]

"Hi @stoic igloo ! 👋
I just opened an issue to stop tracking the chroma.sqlite3 file so we stop getting merge conflicts.

But, Does the current database file contain any essential seed data? If not, I can submit a PR to remove it and update the .gitignore right now. Link: "https://github.com/AOSSIE-Org/BabyNest/issues/173#issue-3848639495"

[uneven linden]

Anybody having any suggestions on this one....

[uneven linden]

@lethal socket
i'm starting on the Intelligence Layer refactor for #151. Basically cleaning up RAGService.js so it can actually handle complex user inputs properly.

@scenic urchin I saw you're debugging the parsing issues in #168. Just a heads-up,since my refactor adds a totally new way to handle inputs (Extractors), you might want to wait until this is merged before you rewrite the chat logic.

otherwise, you'll probably just have to redo your work next week when the structure changes .the new setup should make fixing that bug way easier....
sent friend requests to @lethal socket @scenic urchin

[scenic urchin]

Guys guys, everyone don't waste your time working on the same issue. @strange wasp has already done a PR, once it gets accepted we will work on improving it. If you guys agree we can divide the work like this

1. I will do the "Attaching hooks and make it eligible to call backend endpoints/handlers"
2. @livid drift can make add his OCR thingy and LORA improvement to the model
3. @uneven linden as you are new to this conversation, you can give final touch the pipeline at the end (adding emergency services which skips model for quick call, etc.)

Agree?

[livid drift]

Yeah sure.. currently I am working on OCR feature

[strange wasp]

Yeah, this makes sense.

[uneven linden]

@scenic urchin , appreciate the initiative! But actually, @silent pike and I already spent some time hashing out a specific split for the #151 refactor over on GitHub to avoid exactly what you're worried about,wasted time.

Dheeraj is already on the UI/State/Hooks side (the stuff you mentioned), and I’m taking the lead on the Intelligence Layer (RAGService decomposition). Fixing the RAG logic is a major piece of technical debt that we need to clear out before we can properly solve those parsing bugs in #168. My goal is to get the Intents and Extractors set up so the backend calls actually have clean, structured data to work with.

Let's stick with the plan we already coordinated on the issue thread,it’ll keep us from running into merge conflicts or doing the same work twice. Totally down to sync up once I’ve got the Extractor logic ready for others to use.

[livid drift]

Hi @stoic igloo , I've submitted a fresh Pull Request for the Medical Report Extraction Pipeline with a cleaned-up codebase and generalized assistant naming.

PR Link: https://github.com/AOSSIE-Org/BabyNest/pull/179

Key Updates in this PR:

End-to-End Extraction: Integrated a React Native frontend with a Flask backend for automated vitals scanning.

GGUF-LoRA Pipeline: Implemented local LLM inference using llama-cpp-python, loading a medical-vitals optimized LoRA adapter over a Qwen-0.5B base.

OCR Enhancement: Added an OpenCV preprocessing layer (grayscale, cubic upscaling, adaptive thresholding) to handle noisy report images from mobile cameras.

Standardized Code: Refactored backend logic to use relative pathing and removed all hardcoded system dependencies for better portability.

The pipeline is fully functional and ready for review. I'm looking forward to your feedback on the implementation!

[lone oar]

Hi @stoic igloo and @ivory bolt ! I’ve just opened an architectural proposal to standardize the Appwrite service layer and error handling patterns in BabyNest (Issue #180).

Drawing from my experience with the Resonate architecture, I believe this refactor will greatly improve scalability and developer experience for the upcoming GSoC cycle. I’ve already detailed a proposed solution on GitHub and would love to take the lead on this. Looking forward to your thoughts!

Issue Link: https://github.com/AOSSIE-Org/BabyNest/issues/180

[uneven linden]

@lethal socket @stoic igloo, I think we're about to run into a major merge conflict between #151 (our frontend refactor) and the new #180 (Appwrite refactor).

Right now, both plans are trying to create src/services/apiClient.js and handle the Weight/Health logic, but they point to different places (Python Backend vs. Appwrite).

If we aren't careful, we're going to end up with a situation where the frontend saves weight data to Appwrite, but the AI Agent is looking for it in the Python backend (Postgres).
We should probably decide on the structure now before anyone writes too much code. Maybe we keep apiClient for the Python/AI connection and put the new stuff in services/appwrite/?

[worthy citrus]

@uneven linden You are now level 2! :confetti_ball:

[strange wasp]

Just to clarify, as far as I understand, the current backend uses the Python service with SQLite, and there hasn’t been a maintainer decision yet to move this flow to Appwrite.

Is the Appwrite-based approach something that’s already been approved, or is it still just a proposal being discussed?

[ivory bolt]

Hey @lone oar , I am not mentoring BabyNest

[solid walrus]

heyy i am new here can anyone guide me like do i have to recognize the issues by myself or issues are specified somewhere

[strange wasp]

You can ask the maintainer to assign any unassigned issue to you. To check if an issue is already taken, see the Assignees section or the comments on that issue. If needed, you can also open a new issue to report a problem or request work and ask the maintainer to assign it to you.

[proper musk]

@warm bay 👆

[uneven linden]

understand the project scope and goals and think of what improvements or changes needed to achieve those goals, then raise issues,get assigned and start contributing...
Its very interesting and addictive too😅

[storm heath]

@uneven linden I have a quick question. Is it okay if I get assigned to issues raised by other contributors and start working on them?what should I do to get assign?

[uneven linden]

yeah, its definately nice if you work on issues raised by other contributors😇 😇 , but check if they are already working on them, or failed to complete them! If you work on them directly without checking, it leads to duplicate work 😔 and waste alot time and efforts😢 . But identify and ask(comment) on the issues if you feel like contributing to them. Is this solved your query. Feel free to ask if you have any queries🙂

[strange wasp]

@warm bay @stoic igloo I wanted to share an update that I’ve completed the work for Issue #171. I’ve raised a corresponding PR #182 for it.

While the issue was primarily focused on implementing centralized and environment-aware error handling, during the process I also identified and fixed several related bugs. I additionally separated validators and ensured they generate environment-specific error responses in a consistent manner.

Whenever you get some time, I’d really appreciate it if you could take a look and review PR #182.

Also, if possible, I wanted to gently follow up on PR #170, which is related to fine-tuning the base SLM (Qwen 0.5B). It’s been a bit over two weeks, and I’d be grateful if you could review that as well when time permits.

Thanks a lot! 😊

[uneven linden]

Nice work on #182 . I was just digging into the validation logic myself, nice to see you've got it covered.

[strange wasp]

Thanks! Appreciate you taking a look, glad we’re on the same page with the validation logic

[worthy citrus]

@strange wasp You are now level 6! :confetti_ball:

[worthy citrus]

@livid drift You are now level 5! :confetti_ball:

[livid drift]

Hi @stoic igloo and @warm bay
Hope you're doing well.
I’ve just raised PR #184, which introduces a centralized database service layer to improve the backend architecture. Also, I have an earlier PR #179 pending review.
Whenever you get some time, please do have a look. I'd appreciate your feedback to ensure I'm heading in the right direction with these architectural changes.
No rush, thank you!🙂
https://github.com/AOSSIE-Org/BabyNest/pull/184

[uneven linden]

Hi @warm bay @stoic igloo @proper musk this is regarding the frontend refactor (#151).

A few of us ( @indigo cedar , Vikas, and myself) have been chatting on the issue and we’ve got a solid plan to split up the work so that work can move on smooth.
Basically, we're looking to move toward that 5-layer architecture . @indigo cedar is jumping on the API/Hooks side, and I’m working on the Intelligence Layer (breaking down that massive RAGService.js) since it’s pretty tied into the backend I've been working at.

Before we go all in and start opening a bunch of PRs, could you guys take a quick look at the plan on the issue? Just want to make sure you're agree with the folder structure and the general direction. It’s a pretty big shift for the frontend, so we’d rather get your consent'now than have to redo a alot of logic and code afterwards.
So,Please visit the Issue and give your reviews. treat this as urgent..Hoping you do the same....
Thanks!

[strange wasp]

@stoic igloo @warm bay I had a quick question regarding GSoC alignment.

I’ve been contributing to this project over the past couple of months and I noticed that the project isn’t yet listed on the official GSoC idea list.

As per the admin’s guidance, proposals that align with the organization’s general themes can still be considered even if they’re not explicitly listed, provided they’re discussed publicly first.

So I just wanted to ask, would it be okay to discuss potential GSoC proposal ideas for BabyNest here?

Just looking for guidance on how best to proceed. Thanks!

[lethal socket]

https://github.com/AOSSIE-Org/BabyNest/pull/186
@stoic igloo
@warm bay
I have implemented Notification Service .
please see it and give your feedback on it.

[late lynx]

Hi, I'm Janhvi , sophomore cs major, would love to contribute to this project!

[indigo cedar]

Hi maintainers,I’m working on issue #151 and have started implementing a centralized API client and hooks under src/api/ and src/hooks/ as part of the proposed 5‑layer architecture.Before going further, I’d like to confirm that this direction aligns with the project’s vision.@sharma‑sugurthi and I are coordinating on this and would appreciate your feedback on:the proposed structure (src/api/, src/hooks/, and splitting RAGService.js into domain‑driven modules),whether any parts of the plan should be adjusted.This will help us avoid redoing work later and keep the refactoring aligned with the project’s goals.Thanks!

[stoic igloo]

Hey all!
Please go through this and align it with your proposals
https://github.com/AOSSIE-Org/Info/blob/main/GSoC-Ideas/2026/index.md

[lethal socket]

@strange wasp
Great Work...👍

[strange wasp]

Thanks. I hope it’s reviewed and merged.

[lethal socket]

Hi @stoic igloo
I have reviewed the PR.
I’ve also run and tested the code locally on my system.

Please take a look at the feedback I’ve provided in the PR.

[uneven linden]

hey @stoic igloo ! hope you're doing good..
just wanted to drop a quick msg about the PRs i submitted for BabyNest,, whenever you get a chance to look:

Auth Cleanup: i removed the old @require_auth decorators and auth.py.. felt like dead code since we're going offline-first and it was just cluttering things up.

RAG Refactor: split the RAGService.js to decouple intents/extractors.. makes it way easier to manage the ai logic moving forward.

links are in the repo.. no rush at all, just wanted to ping you so they dont get lost! let me know if any changes needed 🙂

[vestal summit]

Hi everyone 👋

I’m Swathi, currently exploring open-source and preparing for GSoC 2026.

I’ve started contributing to BabyNest and am excited to keep learning and improving through real contributions.

Glad to be here 🙂

[stoic flame]

Is BabyNest not on GSOC 2026 Idea list?

[strange wasp]

This project is not there under individual projects. You can check whether it falls under any of the org wide ideas.

[stoic flame]

I didn't find this in org wide ideas too

Not on AOSSIE website, cuz mayb it's not updated

[teal geyser]

Sir @stoic igloo can you look pr when you are free as I add YouTube channel link in repo
https://github.com/AOSSIE-Org/BabyNest/pull/192

[teal geyser]

Sir @lethal socket @stoic igloo can you review it again I done that alternative change

[stoic igloo]

#1339225155481763882 to avoid rework and maintain decorum, lets not start working on issues unless assigned
Any PRs created without assignment would be closed without review.

[strange wasp]

Hi @stoic igloo , I just wanted to clarify something regarding the PR assignment rule. I had raised a few PRs before this guideline was communicated, and at that time it wasn’t specified that issues needed to be assigned before working on them.

These PRs don’t conflict with anyone else’s work, and no one else is currently working on those issues. I also invested a significant amount of time resolving the bot errors and addressing the required changes.

I completely understand and will follow the assignment process going forward. However, I wanted to ask if the PRs that were already raised earlier could still be reviewed, since they were created before this rule was clarified.

Thank you.

[stoic igloo]

@strange wasp
This is common standards of open source contributions
But I am reviewing most of the open PRs as I understand a lot of time has been spent by devs and wouldn't want it to go to waste

[stoic igloo]

#1339225155481763882 Reply to this message, if I have missed to review or re-review any PRs
make sure video based testing evidences are included in all.

pr format - >
pr title - resolves XYZ issue
pr importance - minor issue/major feature ...
pr link -

Issue format - >
issue title -
issue imp -
issue link -

[teal geyser]

https://github.com/AOSSIE-Org/BabyNest/pull/192
Sir could you review it again as I complete the request changes

[strange wasp]

@stoic igloo The following are my missed PR's

1. PR format ->
   PR title - Adjust date extraction rules based on SLM extraction findings
   PR importance - Major feature involving testing of the fine-tuned model
   PR link - https://github.com/AOSSIE-Org/BabyNest/pull/170

Issue format ->
Issue title - Explore fine-tuning 0.5B SLM for appointment extraction
Issue imp - Helps identify how the existing model can be improved
Issue link - https://github.com/AOSSIE-Org/BabyNest/issues/161

2. PR format ->
   PR title - Feature: Centralized Error Handling
   PR importance - Major feature implementing a fully functional centralized error management system that is scalable and environment-aware
   PR link - https://github.com/AOSSIE-Org/BabyNest/pull/182

Issue format ->
Issue title - FEATURE REQUEST: Centralized and Environment-Aware Error Handling
Issue imp - Improves and standardizes error handling across the application
Issue link - https://github.com/AOSSIE-Org/BabyNest/issues/171

Please check it out when you get time.

[stoic igloo]

@strange wasp kudos on the research
it summarizes well how small datasets fine tuning can be insufficient for SLM training. This shows us that FIne tuned SLMs require huge and high quality datasets training

however my suggestion here would be to focus more on the data quality and collection for pregnancy related articles and not on appointment fixes. We want to accurately convey to the users the kind of medication or appointments they need to undergo and to track their progress in the app simply based on their personal data(country, weight, age)
A small SLM + RAG model might help here.

[strange wasp]

Thanks for the suggestion, that direction definitely makes sense. I’ll revisit this and explore the SLM + RAG approach when I get some bandwidth.
For now, I’m focusing on implementing tests for a sustainability-related idea, but I’d love to come back to this once that’s more settled.

[teal geyser]

Hi sir @stoic igloo , I’d like to work on this issue and start implementing a solution. Could you please assign it to me? Thank you!
https://github.com/AOSSIE-Org/BabyNest/issues/194

[vestal summit]

Hi @stoic igloo

I opened a PR for fixing the Appointment modal UI and keyboard scrolling issue and addressed the CodeRabbit suggestions.

Could you please review it when you have time?

PR: https://github.com/AOSSIE-Org/BabyNest/pull/190
Thank you!

[worthy citrus]

@vestal summit You are now level 2! :confetti_ball:

[teal geyser]

@stoic igloo sir could you review it i done this dependency update to make it our repo secure remove 50+ vulnerabilities
https://github.com/AOSSIE-Org/BabyNest/pull/195

[strange wasp]

@stoic igloo Hi, I just noticed that the BabyNest project has been added back to the GSoC idea list. Since I had previously contributed to the repo, I’ve started working on a proposal for it today.

Would it be okay if I shared my proposal drafts here for feedback as I develop them? Your guidance earlier was really helpful, and it would help me refine the scope and direction.

I’m also in the middle of my university exams this week, but I’ll be actively working on the proposal over the next days. Any suggestions from your side would be really appreciated.

[strange wasp]

Hi @stoic igloo I had a couple of questions while going through the BabyNest idea description.

From what I understand, the core medical scheduling should be entirely rule-based so that reminders and timelines remain deterministic and reliable. In that case, would the AI mainly function as a companion interface (for example summarizing upcoming appointments or helping users interact with the app conversationally), rather than being involved directly in generating the medical schedule itself? I just wanted to confirm my understanding of the intended separation between the deterministic scheduling system and the AI features.

I also had another question regarding scope. Since the goal mentioned in the idea is to bring BabyNest closer to a production-ready state (for Play Store/App Store release), would UI refinements or small feature fixes in existing screens also be reasonable to include as part of the proposal, alongside the larger tasks like the scheduling engine and notification system?

I would appreciate your thoughts so I can scope the proposal appropriately

[teal geyser]

Sir @stoic igloo ,could you please review my pr and merge it when you are free
https://github.com/AOSSIE-Org/BabyNest/pull/195

[lethal socket]

https://expo.dev/blog/how-to-run-ai-models-with-react-native-executorch

[worthy citrus]

@lethal socket You are now level 6! :confetti_ball:

[brazen scaffold]

@stoic igloo

Please check this out

as discussed here
:#announcements message
https://github.com/AOSSIE-Org/BabyNest/pull/191

[stoic igloo]

pls follow the format for PR review

[teal geyser]

Sir @stoic igloo please review it when you are free

[scenic urchin]

Congratulations @strange wasp for GSOC26, you are selected for this project right?

[faint nest]

@stoic igloo could you please join nexus discord, i have something to say there.

[strange wasp]

Yeahhh, thanks so much 😊

[teal geyser]

Congratulations @strange wasp

[strange wasp]

Thanks 😊