hereandnowai/master-langgraph-workflows-in-python-20-real-world-agent-projects-by-hereandnow-ai

Each project implements stateful agent graphs using LangGraph's node-and-edge architecture with conditional routing, human-in-the-loop interrupts, and tool integration patterns—leveraging Cerebras' `gemini-2.5-flash` model via LangChain for cost-efficient inference. The collection spans foundational concepts (basic chat, memory, caching) through production patterns (multi-agent orchestration, vector DB retrieval, map-reduce aggregation), with consolidated dependencies and PEP-8 code designed for rapid iteration. Projects integrate external tools (Tavily search, stock APIs, vector databases) and demonstrate domain-specific workflows across finance, HR, marketing, and log analysis.