CIO, technical operator, backend and AI systems

I lead product and technical work on systems that have to hold up in production.

Most of my work sits where product, backend, delivery, and AI problems overlap — money moving through systems, reports teams rely on, workflows that can't afford to fail.

  • 2-hour onboarding cut to 25 min
  • Reporting endpoints from 750ms to 230ms
  • 15+ specialist agents in production

Recent work

Polaris retail operations illustration

Polaris ERP

Batch stock truth, ledger integrity, and tenant-safe operations.

Retail operations platform covering POS, stock, ledgers, supplier workflows, reporting, and tenant-safe APIs.

Obelisk

Multi-tenant agents, retrieval, reliability, and guardrails.

AI-powered marketing platform with LangGraph agents, parallel RAG, and multi-tenant architecture

Anatomia Healthcare

PHI controls, nurse review, escalation paths, and voice follow-up.

Care workflow platform for nurse callbacks, case review, encrypted transcripts, and voice follow-up.

Recent writing

Production War Stories

I Shipped a Race Condition That Double-Charged Customers (And Other War Stories)

What broke in production, what it cost, and the boring fixes that actually held.

LLMs and User Experience

LLMs Can't Save Bad UX

Why most AI features fail when the underlying product problem is still sitting there.

Multi-agent middleware illustration

Multi-Agent LLM Middleware: Lessons from Marketing Accelerant

How a large multi-agent setup stays usable once cost, context, failure handling, and approvals become real.

Learning Through Open Source

The Best Way to Learn a Codebase Is to Break Someone Else's

Why reading other people's production code teaches more than another polished tutorial.