AI Agent development - LangGraph, RAG

Need help fixing and improving an existing AI agent setup. This is urgent, but it is not a huge project. The main thing we need is to clean up the conditional edge routing between specialized agents so the flow works properly and scales better. Tech stack: Python 3.9, LangGraph StateGraph, conditional edge routing, Anthropic Claude claude-sonnet-4-5 via LangChain Anthropic, ChromaDB for RAG, Sentence Transformers all-MiniLM-L6-v2, and Streamlit UI. We need someone who can improve the routing logic between agents, fix the conditional edge flow in LangGraph, make handoff between specialized nodes cleaner, stabilize the current setup, and suggest small improvements that make the system easier to scale. This is not a long research project. We already have a working base. We need someone who can jump in and fix the routing problem quickly. Please include similar LangGraph or agent work you’ve done, plus a short note on how you’d approach conditional routing between specialized agents.