This is me exploring what's next — a role, a serious collaboration, or just a conversation. Fill out the form. Not sure what that looks like in practice? See how it works first.
Many domains, primarily on Wall Street, taught me one thing about myself: what's constant isn't the domain, it's an instinct for finding the best tools and getting the most out of them, plus a compulsion to keep learning whatever a new domain requires to get there.
The specific tools keep changing. What doesn't is the instinct to go find the best one for the job, learn it fast, and get real leverage out of it — whatever that tool happens to be this decade.
That shift shows up most clearly in how I work now. I spent decades on stage performing with other musicians as a team — the player, after years of practice, who could actually hold their part. Lately I've become more like the conductor: still deeply involved, still accountable for every note, but orchestrating something with far more reach than one instrument ever could.
Here's what that looks like, day to day:
Not a sales pitch for a tool — here's exactly how this played out on one real project this year, start to finish. It's the same way I approach everything else I take on.
You tell me what you're actually trying to achieve, in plain language — the same way I've worked with trading desks for decades. Priorities are allowed to change mid-project the way real business thinking does, instead of locking you into a plan from day one.
Nothing gets built on a guess. Before anything moves forward, the idea gets tested against real, live information — not a made-up example designed to look good in a pitch. Telling a genuinely good answer from one that merely sounds confident is a judgment call, and that judgment comes from decades of experience, not from the tool.
Not "looks right" — actually proven to work, the way it will really be used. On the platform I rebuilt this year, Claude wrote every line of code; I didn't write any of it myself. My job was reviewing every decision, catching mistakes, and making sure the result held up under real conditions — more carefully than most of what I've built solo across my career. Moving faster didn't cost quality here — if anything, it came with more of it.
Delivered once it's actually confirmed working, with a clear account of what changed and why. That platform is the proof: the kind of system that normally takes a team of three or four people at least a year to build. Working this way, it was done in three months, with far more capability than the original ever had.
That's the throughline underneath every one of those steps, and underneath my whole career: not a quant, not an AI enthusiast, but someone who sits in the gap between serious technical complexity and the people who have to trust it without checking the work themselves — at Morgan Stanley Prime Brokerage, at Axioma, part of SimCorp, and now with Claude itself.
Everything on this page is necessarily self-reported — client work stays confidential, the same way yours would be if we worked together. Don't take my word for it; use a conversation to ask the harder questions instead.
One more thing to be upfront about: I'm currently in a full-time role. This isn't a freelance pitch — it's me exploring what's next, seriously, not casually.
If that's the seat you need filled, or you just want to talk about the real workflow behind any of this — let's talk ↑.