How the work goes
The diagnostic comes first. Three to six weeks living with the problem before anyone touches a roadmap: talking to the people who actually use the workflow, tracing the data through the systems it lives in, and pressure-testing whatever brief I was handed in week one. Most briefs survive that process in a different shape than they arrived in. That's the work.
Then the build, with the same person responsible. I write the spec, hold the executive cadence, and stay accountable to the metric we set in the diagnostic. The engineering teams doing the implementation are the ones I'm in the standup with. There is no handoff to a delivery group I never met.
Then the system in production. The job isn't done at go-live; it's done when the people the system was built for are using it instead of working around it. That number, for me, sits behind every other one.
The lineage
Two consultancies co-founded across a decade. Solutions-E for hedge funds and trading desks; Cambrian for global banks, six cities, reporting into CAOs and Heads of Enterprise Architecture.
The shape of the work comes out of two firms I co-founded. Solutions-E, in the early 2000s, ran tech advisory inside hedge funds and trading desks. Cambrian, founded in 2007 with ex-Accenture partners, ran program management for digital transformation across global banks (London, Zurich, Tokyo, Dublin), with Bank of America/Merrill and UBS among the clients. The pattern was Engagement Manager / Client Partner: one practitioner accountable end-to-end, the analysis and the delivery on the same desk.
I have run the same shape inside firms since: at Deutsche Asset Management on the SEC 22e4 product and the State Street cloud partnership; at Morgan Stanley on the CCAR Cross Market Risk workflow; at FactSet on the AI strategy for the Entity Intelligence platform. Different employers, same engagement structure.
The shape, in three engagements
FactSet, 2020–2024. AI strategy and product for private markets data. Team of ~12 across NYC, London, Hyderabad.
Four years inside one product area, owning the AI strategy from data architecture through model training to the client interface. Coverage went from 250K to 3.7M entities; BERT accuracy from 20% to 80%. The numbers held up because I stayed tightly aligned with technical leadership: making sure they understood the business problem as well as I did, and that they were excited to be creative inside it. Same person from the original architecture call to the client-facing release notes and the sprint reviews with sales.
Morgan Stanley CCAR, 2017–2020. Cross Market Risk workflow rebuild on a regulatory clock.
The translation problem in regulatory work isn't technical. It's between four groups: the regulator, risk analytics, the risk managers, and risk IT. The job was to keep the right information moving between them long enough to consolidate 30,000 disparate risk entries into a seven-step auditable workflow and pull Scenario Analytics cycle time from 50 days to 31. I shadowed the product lead through the early meetings with risk leaders and risk ICs, then held the spec, the build, and the presentation back to all parties. I also coached two of the teams through their Agile transition while it was happening.
Deutsche Asset Management, 2010–2017. SEC 22e4 regulatory product; State Street cloud partnership.
Two engagements running in parallel inside one role. The 22e4 product met a new SEC liquidity rule on a hard implementation deadline. The State Street partnership took several hundred reports across 700 users and replaced them with a cloud workflow at 70% lower cost for 150 front-office users. Both shipped on deadline, both with the same person on the hook for the architecture, the report analysis and consolidation, and the change management.
What I won't do
A short list. The first two are about how the work has to be structured to be good. The third is about what I am, and not.
No remote-only engagements. The pattern at every firm that has gotten this right involves a partner across the table, in the building, in the room. Distance erodes the translation. I'm based in Manhattan and get out to the Bay Area several times a year to stay current on the thinking and the energy. When the work calls for it, I'll spend six consecutive days on the ground: the environment, the cadence, the cues that don't survive a Zoom call. I'm not willing to be a face on a screen for the duration of an engagement.
No layers between me and the work. The person who scopes the engagement is the person on the hook when the system goes live. That means showing up to the meetings I'm not formally accountable for, and showing up prepared enough to contribute. The dial-in cameo is its own kind of layer.
Not a yes-man. I'm not wired to simply execute. Where I see strength in a direction, I say so; where I don't, I say that too. That has always been the value. The alternative is expensive theater.
If something here resonates with how you're trying to run a program (at a firm, at a fund, at a platform), there's a small number of conversations a year I take outside the current engagement. I read every note that comes in through LinkedIn.