Why not just hire someone?

Hire, if you can. A senior backend or AI engineer takes three to six months to find and then sits on the budget forever. I’m useful in front of that. The system has to exist this quarter, and whoever you hire later inherits something that already runs.

Why you and not an agency?

An agency sells you a team and staffs it with whoever is free that month. You get me, in your repo, for the whole engagement. Fewer people, fewer handovers, and nobody translating your problem to a junior on the way to the commit.

What if it turns out to be the wrong call?

The first scope is small on purpose. You see a demo every week and every commit lands in your repo from day one, so if it’s going wrong you find out in week one. Stopping there costs you one small scope instead of a quarter.

What happens when you leave?

You keep a system your team can run. The runbook, the architecture walkthrough and a handover session are in the scope from the start. Everything lives in your repos on your infrastructure, so nothing breaks when I stop answering.

Backend · AI systems · Full-stack

Systems that hold up when being wrong costs money.

I build trading engines, exchange integrations, and the agent systems teams are rushing into production without evals. Ten years of it. Right now that means a Go hedging engine holding market exposure for a $38B exchange.

See the work

[01]Selected work

Things I built that had to stay up.

All projects

Ostium

Contract senior engineer

A hedging engine holding market exposure for a $38B exchange.

volume: $38B+volume
end to end: <50msend to end
broker integrations: 6+broker integrations

Details →02

QA agent pipeline

Contract

PRDs in, reviewed test plans out. 60% of the cases run without a human.

fewer production bugs: 40%fewer production bugs
fewer regression bugs: 30%fewer regression bugs
cases run without a human: 60%cases run without a human

Details →03

Rage Trade

Contract tech lead

$6B+ in trades routed across 8+ exchanges.

volume routed: $6B+volume routed
exchanges: 8+exchanges
latency cut: 50%latency cut

Details →

[02]What I do

Three problems I get called for.

AI systems that survive production

The demo worked. Month three is where it hurts: quiet regressions, a bill nobody forecast, and no way to prove a change made things better. I fix that with real evals, tighter agent architecture, better retrieval and tool design, and hard caps on cost and latency.

LLM pipelines
evals
agent harnesses
traces
cost caps

Backend and integrations

One upstream goes down mid-request and takes your endpoint with it. I build the high-throughput services and the integration layer underneath: a dozen unreliable providers behind one clean internal contract, with the retries, timeouts and fallbacks for the day one of them dies. It ships to your infrastructure and it stays up.

Go
Python
TypeScript
Postgres
Redis
gRPC
AWS

Full-stack delivery

The plan is agreed. The quarter is committed. The one engineer who could own it already has a job. I take it from architecture to deployed and working, on my own, and hand back something your team can run.

Next.js
React
Node
CI/CD
observability

[03]How it works

01Free
A call
Thirty minutes on your system. Come with the failure you can’t explain, and I’ll tell you what I think is causing it and whether it’s worth scoping.
02Within 48h
A written scope
What I will build, what I will leave alone, how long it takes and what it costs. One page. Priced per engagement, because your problem doesn’t fit a package.
03Included
Build and hand over
Shared channel, a demo every week, every commit in your repo. The runbook and the handover walkthrough are written into the scope, so nobody has to ask for them at the end.

[04]Straight answers

Before you book.

Why not just hire someone?: Hire, if you can. A senior backend or AI engineer takes three to six months to find and then sits on the budget forever. I’m useful in front of that. The system has to exist this quarter, and whoever you hire later inherits something that already runs.
Why you and not an agency?: An agency sells you a team and staffs it with whoever is free that month. You get me, in your repo, for the whole engagement. Fewer people, fewer handovers, and nobody translating your problem to a junior on the way to the commit.
What if it turns out to be the wrong call?: The first scope is small on purpose. You see a demo every week and every commit lands in your repo from day one, so if it’s going wrong you find out in week one. Stopping there costs you one small scope instead of a quarter.
What happens when you leave?: You keep a system your team can run. The runbook, the architecture walkthrough and a handover session are in the scope from the start. Everything lives in your repos on your infrastructure, so nothing breaks when I stop answering.

Thirty minutes, and you’ll know if I’m useful.

No deck, no discovery phase. Bring the thing that keeps breaking. You leave with a straight read on what’s wrong and what it takes to fix, or an honest no.