I design and ship production GenAI systems — agent orchestration, RAG, human-in-the-loop workflows, and cloud deployment — for insurance, document AI, and enterprise use cases.
| Area | What I build |
|---|---|
| GenAI agents | Tool design, instruction engineering, MCP integrations, multi-step reasoning |
| Production systems | FastAPI / ADK APIs, Next.js dashboards, Docker, GCP Cloud Run, IAM |
| RAG & memory | Chroma, LangMem, session persistence, audit-ready tool traces |
| ML engineering | End-to-end pipelines, MLOps, monitoring, model deployment |
Production claims-assistant on Google ADK — custom tools, persistent sessions, Next.js adjuster dashboard, Cloud Run deployment.
Impact: Faster FNOL draft generation · consistent policy-grounded answers · full tool-trace audit path
Stack: Google ADK · Gemini · Tavily MCP · Next.js · Docker · GCP Cloud Run
Human-in-the-loop insurance FNOL copilot — LangChain agents, LangMem, Chroma RAG, FastAPI, Next.js.
Impact: Adjuster-assisted triage · coverage recommendations grounded in policy KB · safe rollout by design
Stack: LangChain · LangMem · Chroma · FastAPI · Next.js
Reusable ML / data-science playbooks I maintain as open source — from problem framing through feature selection.
Generative AI & agents
Cloud, MLOps & data
Machine learning & analytics
Open to consulting, GenAI architecture reviews, and production agent builds.
