Skip to content

About

In 2004 I inherited a hand-me-down Pentium III that switched itself off whenever it got hot. I installed Linux on it and nothing worked. Not the display, not the sound, not the dial-up. I copied error messages onto paper by hand, carried the paper to an internet café, and paid by the hour to search for answers I could carry back.

The network card, I never fixed. That one beat me.

Two decades on, I work on clocks that live on network cards. At Gemini I put PTP on Solarflare hardware so that every timestamp beneath matching, surveillance, and the audit trail was honest to the nanosecond. When a regulator or an auditor or one of your own engineers asks what happened and in what order, the answer has to be true. The clock is where that truth either starts, or quietly stops.

Most of that work is invisible when it goes well. Nobody celebrates the failover that held, the backup that restored, the escalation that woke the right person at 3am. My career is a list of disasters that never happened.

So I build for the version of me that is tired and wrong. At Upside, every mainnet deployment ran as a witnessed release: a written pre-flight checklist, a full testnet rehearsal, and a second and third pair of eyes on the console before anything touched real money. Sixteen months, more than half a billion dollars in custody, no incidents. That record is not a story about being careful. It is a story about assuming I will not be.

The machines are only half of it. The other half is whether an engineer’s hesitation gets heard, whether bad news travels upward fast enough, whether an organisation keeps a spare for its best judgment the way it keeps a spare for its best server. That is what I spoke about on the TEDxCMRIT Hyderabad stage.

Now

I work independently. I’m building ZeroCopy Systems, confidential-computing signing infrastructure for institutions that cannot outsource custody of their own keys, and I take on a small number of engagements and speaking dates each year with teams running systems that move money.

What interests me most right now is the infrastructure underneath autonomous agents that touch capital. We are about to hand real money to software that decides on its own, and the controls around it are still mostly documents. They need to be hardware.

Remote-first with teams in the US, Europe, the UAE, and India, on-site when it matters. No service catalog. Scope is a conversation.

Where I’ve worked

  • 2025 to now

    Founder, ZeroCopy Systems

    Confidential-computing signing for institutions that cannot hand their keys to a vendor. Keys are generated inside the enclave and never cross the boundary. Every build is reproducible, so a counterparty can check what is running without taking my word for it. Pre-launch.

  • 2022 to 2024

    Founding Web3 Infrastructure Engineer

    Upside. I owned the wallet infrastructure holding more than $500M across hot, warm, and cold tiers, on HSM-backed signing with multi-sig approvals and audit-grade logging. Sixteen months, zero security incidents. Every mainnet deployment ran as a witnessed release: written pre-flight checklist, full testnet rehearsal, and a second and third pair of eyes on the console. The mistake I cannot catch myself is the one worth designing around.

  • 2021 to 2022

    Founding Cryptocurrency DevOps Engineer

    Akuna Capital. I built the crypto desk’s trading infrastructure from nothing, across twelve venues. To find the fastest seat inside each exchange’s availability zone I booted hundreds of bare-metal instances, benchmarked every one against the live venue endpoint, kept the quietest one percent as production hosts, and destroyed the rest. Placement you can measure beats placement you can argue about.

  • 2020 to 2021

    Senior Site Reliability Engineer

    Gemini. Exchange platform team, holding matching, market data, and order entry above 99.99% uptime on $2B+ of daily volume. I put PTP on Solarflare hardware so the timestamps under matching, surveillance, and the audit trail were honest to the nanosecond. An exchange is only as fair as its clock.

  • 2019 to 2020

    Site Reliability Engineer

    Pex. Content identification across billions of audio and video events. I moved the platform from GCP onto Azure Kubernetes and took 45% out of the running cost, then built the monitoring that caught most failures before a user ever felt one.

  • 2012 to 2018

    Lead Linux Infrastructure Engineer

    FlexTrade. Fifteen hundred production Linux systems across five sites, under banks, hedge funds, and prop desks. I led a team of four and built the SOC 2 and MiFID II frameworks the platform was audited against. We passed every audit. Six years is long enough to watch your own shortcuts come back for you.

Client names inside these engagements stay private. Anything I state as a number, I can show you how it was measured.

If you want the disasters at your company to stay hypothetical, let’s talk.

Reach out on LinkedIn ↗