Consulting

I take on a small number of engagements a year. The work I'm best at sits where hardware meets software: getting trustworthy data off a device, building the distributed systems and AWS infrastructure around it, and putting agentic software on top with verification that actually bites. I've done this as an owner at Amazon, as a sole founder shipping an AI product, and now inside a medical-device company.

• Sensor data and physical-AI infrastructure

  From on-device acquisition to the cloud: signal pipelines, storage you can trust after the session, observability, and the study or fleet tooling that sits on top. Wearables, ultrasound and biopotential systems are home ground.

• Distributed systems on AWS

  Event-driven services, real-time data delivery, and the operational work that keeps them boring: CI/CD, infrastructure as code, cost, and on-call reality. Four years of this at Amazon scale.

• Agentic pipelines with real verification

  Multi-agent workflows with review gates, specs as the source of truth, and test harnesses that check what the agents did. I run my own orchestrator in production and can design, audit or build yours.

• Zero-to-one product engineering

  A mobile or web product from a blank repo to a beta in the field: React Native, Next.js, Python backends, LLM features done properly. I shipped Selver this way, alone.

• Advisory

  A few hours a month. Architecture reviews, hiring help, a second opinion before a big decision.

• Embedded

  One to three days a week inside your team for a quarter, owning a system end to end.

• Build

  A scoped deliverable with a fixed end: a pipeline, a platform, a proof of concept that has to be real.

• I write code and own outcomes; I'm not a slide-deck consultant.
• Everything I ship is observable, tested and documented, because I'll be the one woken up by it otherwise.
• Clear scope, clear price, honest about what I don't know.