At Zendesk, I built the backend and platform foundation for production AI agents that take actions across Zendesk and connected systems. I architected tenant-isolated knowledge retrieval on AWS, grounding agents in Guide content, connectors, ticket history, and live CRM and order APIs.
I led Python ingestion and retrieval services, built Go and Rails tool-execution services, and delivered Zendesk as an MCP server with OAuth and per-tenant tools. I also extended MCP client capabilities, integrated Amazon Bedrock, and added Langfuse observability for agent traces, prompts, cost, and quality.
Before AI agents, I delivered shared S3-backed file handling, Kafka and Protobuf audit streams, and enterprise audit-log search prototypes across Zendesk products. At ATHN, I led the full-stack platform behind a national patient registry used by more than 140 specialty clinics, improving study setup from months of custom work to reusable workflows delivered in weeks.
I've built full-stack SaaS products at Invoca, ATHN, and Zendesk across Rails, React, Go, Python, AWS, Kubernetes, and PostgreSQL. I bring 11+ years of experience building reliable systems in compliance-heavy environments including SOC 2, HIPAA, and SOX.
