Production LLM systems
Multi-agent workflows, RAG with explicit grounding and hallucination guards, structured tool-use — built with the operational discipline that keeps them working past the first demo.
Senior engineer · LLM applications
I design and ship multi-agent LLM workflows, governed AI collaboration systems, and the backend platforms underneath them — with the operational discipline that keeps real systems working past the demo.
Focus areas
Positioning
Multi-agent workflows, RAG with explicit grounding and hallucination guards, structured tool-use — built with the operational discipline that keeps them working past the first demo.
Memory and context discipline, human approval gates, structured handoffs between specialist agents, auditable batch protocols. The opposite of autonomy theater.
Seventeen years of payments, messaging, and platform modernization underneath the AI work — so the system underneath the model actually holds up.
Selected work
Role: Architect & Builder
System: Multi-tenant content product for B2B hospitality · object-model law, tenant and property isolation
A multi-tenant content product for B2B hospitality, built behind a governed multi-agent operating system, with a hard ob…
Role: Architecture / Backend Lead
System: Lifecycle messaging workflow orchestration system · idempotency, concurrency control
A flow-based backend system for CRM-driven WhatsApp lifecycle messaging, built with queues, scheduling, tracking, idempo…
Role: Architecture / Backend Lead
System: Outgoing payments reporting and operations platform · idempotency, locks
A business-critical outgoing payments and reporting system that unified multiple payment sources behind one admin surfac…
Writing
An AI job-triage tool I open-sourced this week, and the case for designing AI tools around what they refuse to…
My classifier flickered on identical input — so I voted harder, like the last post taught me. It kept flickeri…