Philip Simon Derock

At Blackcoat AI, I engineered production-grade Agentic RAG infrastructure for legal research, orchestrating LangGraph ReAct multi-agent pipelines across legal contracts and acts. I served the system through FastAPI and FastMCP so external AI agents could use legal intelligence as MCP-native tools.

I've built Lunarbit, a finance-first GraphRAG product that transforms six years of commerce data into an auditable Neo4j knowledge graph with 48K+ nodes and 69K+ relationships. I also built DEX Jobs, a zero-cost AI job-hunting agent with eight specialized agents, multi-tier memory, dynamic LLM routing, and a concurrent pipeline across 19 job sources.

My work spans fine-tuning governed clinical-reasoning AI, AI coding copilots, document workflow automation, healthcare prediction, and applied AI research. I work primarily with Python, LangGraph, LangChain, FastAPI, vector databases, knowledge graphs, retrieval systems, and LLM alignment methods.