Inspiration
We all need it or know someone in dire need of first aid. In developing countries good first aid means the difference between life and death as access to healthcare professionals. Conducting online search on how to do first aid is an uphill task and time consuming for an average person and information
What it does
To address the challenges faced, i developed FirstAider as a first aid information chat-bot to:
- Help people access to up-to-date information.
- Save time by collecting and presenting all information inform of a chat application.
- Make sure all information is vetted by healthcare professionals.
- Make information easy to use and access.
How we built it
- Langchain fro document and web loaders, retrievers, embedding models, vector stores and chat models
- MongoDb Vector Search Used to store and efficiently retrieve content as vectors for Retrieval Augmented Generation (RAG), enabling powerful and accurate chat responses.
- MongoDb as a database.
- Python Flask as API framework.
- Angular for UI/UX development.
- Vercel for front-end hosting.
- Google Cloud for backend/API hosting.
Challenges we ran into
FirstAider being a health related chat-bot, getting verified content was a challenge. Using Langchain document loaders i.e pdfs with a web service was challenging seeing Langchain uses file path which is impossible when using web services.
Lessons
I have learnt a lot while developing this project from Langchain to MongoDb Vector Search . Seamless integration of all this technologies from vector embeddings, document loaders, chat models e.t.c with a blend of simple user friendly chat-bot design.
- Langchain
- MongoDb
- MongoDb Atlas Vector search
- Angular
What's next for First Aider
I will continue to develop and fine-tune FirstAider. My vision is for it to become the go-to tool for First Aid related information, by continuously updating the knowledge base and improving the AI capabilities of the chat-bot.
Log in or sign up for Devpost to join the conversation.