Implementing a Multi-Tenant Generative AI Chatbot.
summary
Live demo shows a working multi-tenant RAG chatbot deployed to EKS that uses LangChain + a FastAPI “rag API”, Amazon Bedrock for models (Anthropic Claude for generation, Titan for embeddings), DynamoDB for session/chat history, Streamlit UI, and FAISS as a local vector store per pod. Tenants are isolated by Kubernetes namespaces with separate indexed PDFs so one tenant cannot answer questions about the other. All code and a reproducible GitHub repo exist. multi-tenant isolation Bedrock + LangChain FAISS local vectors
AI-written summary · about
chapters (8)
- 0:00Introduction and overview
- 1:01Multi-tenant chatbot demo
- 7:15Generative AI and RAG basics
- 9:59Bedrock and model selection
- 12:23LangChain framework deep dive
- 18:19Prompt engineering strategies
- 24:16Embeddings and vector storage
- 27:02Architecture and final thoughts
related talks
26mReproducible AI with Kubeflow, lakeFS and Langchain
38mGenerative AI - Architectures and applications in depth
27mFrom Chatbots to Agents: Orchestrating Agentic AI on Existing Enterprise Kubernetes Infrastructure.
39mPowering Generative AI with Kubernetes: A Cloud Native Approach
27mGen-AI at Scale: Simplifying Orchestration of Healthcare Applications Across Multi-Cluster Environme
33mScalable LLM Inference on Kubernetes With NVIDIA NIMS, LangChain, Milvus and Flu...
31mMachine Learning-Enabled Chatbots
16mEnterprise AI Innovation: Gen AI Experimentation