Co-pilot for the web.
The project is divided in 3 modules.
- Frontend : React.js frontend using the Ant Design component library.
- Backend : Django backend using Redis for KV store.
- Inference : FastAPI server running an inference endpoint for RAG on GPT4All's llama-2-7b-chat.ggmlv3.q4_0 powered by Langchain. Also performs data scraping using BeautifulSoup and caches the embeddings into a persistent Chroma vectorstore.
Built by Sarang and John for the Moveworks Generative AI Hack-AI-thon 2023 at BITS Goa.