Skill

LangChain

Framework of composable LLM building blocks — models, prompts, tools and retrievers — that I use under my LangGraph labs.

  • #ai
  • #framework

LangChain is an open-source framework (Python and JavaScript) that provides composable abstractions for chat models, prompts, tools, retrievers and output parsers. In my work it is mostly the layer underneath LangGraph.

How I use it

  • langgraph-langserve-starter — swaps between LangChain chat model integrations (Gemini, OpenAI, Anthropic) with a one-line switch and falls back to a fake LLM for zero-key runs and tests.
  • local-rag-chatbot — LangChain loaders, recursive chunking and a Chroma vector store feed a LangGraph retrieval graph.
  • langgraph-ts-agents — LangChain.js withStructuredOutput with Zod schemas drives supervisor routing; tests use a fake chat model so they run offline.
  • mcp-deep-dive deliberately avoids LangChain adapters to expose the raw MCP (Model Context Protocol) protocol — a conscious trade-off I document.

Credentials and courses

LangChain is central to the Agentic AI Engineering Specialization specialization (Building Your First AI Agent with LangChain, Applied Agentic AI Pipelines with LangChain) and to the IBM Generative AI Engineering Professional Certificate certificate (AI Agents Using RAG and LangChain, Generative AI Applications Specialist).

Related