Mehzar Kamzi

At JPMorgan Chase, I lead architecture and hands-on delivery for production generative AI products, agentic workflows, secure APIs, retrieval systems, and multi-cloud AI services in regulated financial environments.

I've built Python-based LLM workflows with LangGraph, Google ADK, FastAPI, and Pydantic, taking prototypes into production with tool use, structured outputs, human review, resilient state handling, and roughly 24% higher scoped task completion. I also design multi-agent systems, governed RAG, evaluation suites, and production instrumentation for quality, latency, cost, reliability, and traceability.

Previously at Verizon and Concentra, I delivered distributed Java, Python, TypeScript, and React applications, event-driven services, secure healthcare workflows, and data-processing systems. My work improved recurring integration diagnosis by approximately 18% and reduced manual data-preparation effort by about 25%.

Across 15+ years, I've turned ambiguous data, integration, and product requirements into dependable software through architecture ownership, mentoring, security controls, automated testing, and practical collaboration with engineering, analytics, and delivery teams.