Build an Intelligent Python App, No Infra Hassles!
summary
A concrete RAG chat app gets built and demoed end to end: Python code uses LangChain + Chainlit UI, FastAPI, Hugging Face embeddings and a Quadrant vector DB, with documents chunked and indexed via a local script; Azure OpenAI supplies the LLM and Azure Container Apps hosts the web app and a file-indexing job, mounting Azure File Share and using the Quadrant vector DB add-on. Deployment uses container images or az container app up, identities replace API keys, the app works but shows typical LLM rough edges (wrong groupby and occasional tooling hallucinations), and Dynamic sessions for secure Python REPL run code remotely with low latency.
AI-written summary · about
chapters (11)
- 0:00Welcome
- 2:30Building RAG app locally
- 5:40Local setup demo
- 15:55Chunk size for RAG
- 16:16Deploy to Azure Container Apps
- 20:15Cloud architecture
- 30:01Extend with Python REPL
- 41:20LLM code execution risks
- 43:05Performance and lifecycle
- 44:15Dynamic Sessions intro
- 51:11Dynamic Sessions benefits
related talks
16mBuild a RAG app in minutes using Langflow OpenAI and Azure
44mThe RAG Really Ties the App Together
27mFrom RAGs to riches: Build an AI document interrogation app in 30 mins
41mFrom RAGs to riches: Build an AI document interrogation app in 30 mins
23mBuild intelligent apps with JS: integrating RAG, Azure OpenAI and Langchain.js
38mQuerying Live Data With LLM App
55mBuilding and scaling AI applications for Data Scientists and Developers
28mGenerative AI-powered app development