How we can work together

Hands-on evaluation engineering for AI agents. You can hire me into your team, or have me review a system you already run.

Embedded AI engineer

I join your team and ship features along with the evaluation pipeline that measures them, with a weekly written status.

  • Before a feature ships, it has a dataset, defined metrics and a release gate.
  • Tracing with OpenTelemetry and Arize Phoenix, so regressions are visible, not felt.

Good fit if

  • You build AI where a wrong answer costs an expert's time or a customer's trust.
  • You have an LLM or voice feature in or near production.
  • You want hands-on engineering, not a slide deck.

Architecture and eval review

The proposed review covers one agent system: evidence-linked failures, evaluation gaps, prioritized fixes and explicit limitations; implementation is scoped separately.

  • Failure modes, with examples.
  • Eval gaps, and the datasets or metrics that close them.
  • Fixes in priority order, so your team knows what to do first.

Good fit if

  • You ship an LLM agent to customers and need to understand where it fails.
  • Your system is in or near production.
  • You want an engineer's written findings your own team can act on.

Hands-on engineering only, not team-lead or manager roles.

Rates depend on scope and committed weekly hours.

How an engagement starts

  1. 1A 30-minute discovery call

    Tell me what you're building, what's failing and when you need help. We decide whether an engagement or a review fits.

  2. 2A written scope

    I send what I'll build or review, the committed weekly hours and the rate.

Start with a discovery call

Pick a time below, or open the calendar in a new tab.

Book a 30-minute discovery call

Prefer email? Send a note.