About - One principal. A fleet of agents.
ESARC is a small, senior-only AI engineering shop. I run it. I built it because the math on AI-native teams changed last year, and most consultancies haven’t noticed yet.
I started ESARC after a year of watching AI agents do the work that used to take a full team.
I’ve been an engineer for 9 years. Amazon for four (shipping pricing services at 1M requests a day, $100K saved on a data map rewrite, half the latency on the cart pipeline). Scrubs Co-Pilot for two (RAG over EHR data, EPIC integration, live in 5+ clinics). Stuf Storage for one (Sidney Voice AI, an autonomous sales agent that books leads on inbound calls).
Today I’m embedded with Meta’s AI Pathfinding team inside Meta Superintelligence Labs. Tool-using agents on Llama and Claude. A multi-agent pipeline that writes, tests, and iterates VR game logic from a product spec.
Working at that level taught me something the consulting world hasn’t caught up to yet: one senior engineer with a fleet of well-orchestrated agents now ships what a team of six used to ship. The economics break the Big-4 model entirely.
That’s what ESARC is. Senior engineer in your repo, agent fleet doing the rest, output on the order of a much bigger team. I take a small number of engagements a year so each one gets the real attention. See the recent work for the receipts.
How we operate
Five rules I won’t bend on.
01
Senior-only
Every line that ships is reviewed by a principal. No junior pyramid, no offshore relay, no learning curve on your dollar.
02
AI-native
Agents draft, test, and review. The principal sets direction and owns the call. Throughput like a team of ten, accountability of one.
03
Receipts over slides
You get pull requests, not status decks. Every claim has a commit. Every milestone has a working build.
04
Async-first
Loom over Zoom. Notion over meetings. Standups are a daily PR list, not a calendar block. Your team keeps its day.
05
Clean handoff
Runbook, evals, on-call notes. When we leave, your team owns it on day one. We stay on call for 30 days, then we’re out.
Who we’ve shipped for
Companies that trusted the principal-led model, from billion-user systems at Amazon and Meta to early-stage AI builds at Stuf and Scrubs.
Tell us what you’re trying to ship.
A principal engineer reads every inbound. We reply same day on weekdays, with an honest read of whether we’re the right team for the work.