Consulting

I help senior engineering teams ship reliable multi-agent systems in Python.

No fake-green test theatre, no agent demos that fall over on second use. Forward-deployed engineering on your repo, on your stack, with a deterministic orchestrator (Bernstein) underneath so the work is reproducible after I leave.

Things shipped

  • Bernstein — open-source multi-agent orchestrator. 50+ CLI adapters, Apache-2.0, deterministic scheduling, HMAC-chained audit trail.
  • Grounded legal QA placed 38th of 356 in a public AI competition. Page-level citations, hybrid retrieval, evidence-first answers.
  • Marketing operations platform built solo in three months. Next.js + FastAPI + ClickHouse unifying seven ad platforms and four revenue partners behind an agentic advisor with three-tier tool-calling.

Office hours

30 minutes, no pitch. For senior engineering leaders working on agent infra, RAG, or eval-driven delivery.

Two-line intake so I can prepare: a one-line problem description, plus team size and domain.

What working with me looks like

  • Architecture review — one week. For teams with a near-live AI system that needs clearer seams and less accidental chaos.
  • Bernstein deployment — two to four weeks. Stand up a multi-agent crew on your repo with file-based state, per-agent credential scoping, signed audit trail.
  • Embedded build shaping — two to three months. Senior engineer on the team to shape workflow, system boundaries, and path to production.

Case studies

Redacted write-ups from prior engagements are shared on request after a first call.

Rate band

Engagements are scoped to a problem and priced on outcome, not seat-time. The default shapes are fixed-bid or capped time-and-materials. I share a concrete number on the first call after I read your intake.

Reading before the call