class PrathikRavichandran:
role = "Agentic AI Systems Engineer @ Medical Metrics, Inc."
experience = "4+ years"
focus = ["Claude Code primitives", "MCP servers", "Multi-agent orchestration",
"Hybrid RAG", "HIPAA / FDA-regulated AI"]
daily_tool = "Claude Code — Skills, sub-agents, hooks, plugins, MCP, settings"
building = "Autonomous agents that think, plan, and act — under audit"
location = "Houston, TX"
belief = "Reliability beats capability. Guardrails are a feature."
🛤️ pathwaysConversational AI navigator for post-incarceration reentry in Texas. Claude Code architecture: layered Skills, sub-agents, deterministic hooks, MCP servers — composed into one auditable workflow.
|
Multi-agent job matching: resume parser → job searcher → match scorer, wired through LangGraph. FastAPI backend with split deploy across HF Spaces and Vercel.
|
|
Autonomous email-to-appointment pipeline. 9 isolated FastAPI agents over an A2A protocol, strict state machine, Postgres audit log, Slack HITL gate, Gmail/Calendar integration.
|
🏭 PlantOpsPlant operator Q&A. Hybrid ChromaDB + BM25 retrieval, RRF fusion, distance fallback, cited PDF previews. Routes every question to the right manual and facility before answering.
|
🏠 PropAIProperty-manager AI agent. Real-estate portfolio dashboard with tenants, loans, escrow, maintenance — and Gemini-powered insights grounded in the live data.
|
Currently building production multi-agent systems for Medical Metrics, Inc. under HIPAA, ISO 27001, and 21 CFR Part 11 — MCP servers connecting an on-prem LLM to clinical databases for a 60-person clinical org.
|
-7b2fff?style=for-the-badge&logoColor=white){kind=icon}
Open to senior agentic-AI / LLM systems roles. F-1 OPT authorized through early 2029; long-term sponsorship welcome.
"Reliability beats capability. Guardrails are a feature."