muhammad.yousaf
All work
2024Project ManagerClosed-source

Benjitron-9000

LangGraph-powered AI agent with human-in-the-loop automation and persistent memory.

FastAPILangGraphAzure OpenAISQLModel

Summary

Agent-based workflow automation where humans approve or correct mid-flight AI decisions. Persistent memory across runs, tool-use through typed interfaces, and clean separation between deterministic business logic and LLM reasoning.

Highlights

  • LangGraph state machine with interrupt + resume at human review gates
  • SQLModel-backed memory store to keep agent context across sessions
  • Deterministic guardrails wrapping the LLM tool calls
This project is closed-source (built for a Kcube AI client). I'm happy to walk through the architecture, trade-offs, and code on request.