Skip to content

For Prop Shops · Crypto Funds · Trading Desks

Building AI agents instead? →
Accepting 1 new Fractional · June 2026
30-40 hrs/wk contract · Q2-Q3 2026

Your signing path has a tax. Most shops have never measured it.

I built crypto infra at Akuna Capital and SRE'd at Gemini Exchange. I've worked at firms that measure in microseconds, and audited firms that didn't know their p99. I find the latency. I fix it.

Your signing path, your key management, your execution loop. Each one has a cost no one's quantified yet.

42µs signing latency 15 years in production 40+ Rust crates shipped Ex-Gemini · Ex-Akuna

Why this engagement is different.

I didn't advise on this. I built it.

I've built a TEE signing platform at 42µs

Most fractional CTOs advise on infrastructure. I've built it. ZeroCopy's AWS Nitro enclave is the reference implementation you'd otherwise hire a 3-month consulting firm to scope and design.

Both sides of the latency gap

Akuna Capital + Gemini Exchange

Prop shop crypto infra at Akuna. Exchange-scale SRE at Gemini. I know what institutional-grade signing looks like because I built and ran it. Not because I read the whitepaper.

AI does the parallel work. I own the judgment calls.

Automated where safe. Human where it matters.

Code review, test generation, dependency scanning, documentation: automated. Architecture, security, anything where being wrong is expensive: that’s mine.

Where your alpha leaks.

Every millisecond between signal and signed transaction is alpha you don't capture. Most shops don't know where the time goes.

Key management

AWS KMS: 130ms · CloudHSM: 5ms · TEE: 42µs

3,095× faster signing at the hardware level. The delta compounds across every trade in your book.

Signing path p99

Typical shop: 8-40ms · Best-in-class: <500µs

Most shops have never measured their signing p99 in production under load. The number is almost always worse than expected.

Order execution loop

strace overhead: 166× · eBPF overhead: <500ns

If you're profiling with the wrong tools, you're measuring the profiler. I know which tool fits each layer of the stack.

First-week deliverable: the latency audit.

. Signing path p50/p95/p99 under realistic load
. Key management architecture (HSM, KMS, TEE, software)
. Order execution loop timing. Signal to signed submission
. Circuit breaker and kill-switch latency overhead
. NATS / messaging bus jitter profile
. Linux kernel settings (THP, CPU governor, C-states, IRQ affinity)
. Network path. Co-location, cross-DC hops, VPN tax
. Exchange adapter overhead per venue

Output: written report with severity ranking and remediation sequence. Most shops find 2-3 items in the top tier that they can fix in a week.

Track record

Gemini Exchange. SRE

Exchange-scale reliability engineering. Know what institutional-grade signing infrastructure looks like in production at real transaction volume.

Akuna Capital. Crypto Infra

Prop shop crypto infrastructure. Know what the latency bar looks like at a firm that trades for a living. That bar is the benchmark I use for audits.

ZeroCopy Systems. TEE Platform

Built a production TEE signing platform from scratch. 42µs deterministic signing. 40+ Rust crates. This is the implementation reference.

Quantfund Engine

Built the systematic trading infrastructure: kill-switch, circuit breaker, Telegram alerts, paper-to-shadow-mode pipeline. 1,000+ tests. Shipped infrastructure.

Free · No call required

The Latency Audit Template. 8 checks most shops have never measured

  • . Signing p99 baseline across hardware security module vs. KMS
  • . KMS API overhead attribution (network, serialisation, queue)
  • . Linux kernel tuning checklist for sub-millisecond determinism
  • . Exchange adapter round-trip decomposition (encode → sign → transmit)

How to engage.

Most prop-shop engagements start with a scoping call to determine the right intensity. Advisory works for a one-shot latency audit and remediation plan.

Engagement framework + retainer modes →

What's your current signing p99?

If you don't know the number, that's where we start. Three questions, reply within 24 hours. Or book directly: calendar.app.google/W1CEqo8GsoGtjJX49

e.g. "Series A fintech, $8M raised 2024, 12-person team"

What keeps you up at night? Be direct. Vague answers get vague responses.

No sales pitch. One conversation to see if there's a fit.