How I work

I decide what matters in sprawling Tech portfolios, then lead the teams that deliver it.

Portfolio management to choose the work that drives the most impact, program management to deliver efficiently, team leadership to grow the people doing it.

01 · PORTFOLIO

Start with impact, not activity.

Thirty-plus solutions don't all matter equally. My job is knowing which five do this quarter, and protecting my teams' time for them. On a six-month stretch into Nike's analytics org I re-pointed the whole backlog at leadership's strategic priorities; once we stopped doing low-value work, other teams stopped wasting time on it too. Ruthless prioritization is the most underrated form of respect for the people doing the work.

02 · TEAM

Lead the people, not the plan.

I grow TPMs, engineers, testers, and BSAs, connect them to work that stretches them, then get out of the way. I've led teams of 5 direct and 6 extended reports the same way: Clear context, real trust, no hovering.

03 · PORTFOLIO

Make the portfolio legible.

Leaders shouldn't have to dig for status. I build the processes and visibility that turn 30 workstreams into clear decisions. That work cut time-to-prioritization by 20% and kept leadership focused on the strategic objectives that count.

04 · PROGRAM

Bet on the new thing early.

The GenAI rollout started as a hunch about where contract work was headed. I drove it across five teams covering generation, search, and review, and it shipped with projected annual savings in the eight figures. It wasn't the first time: At Roche I wrote the business case for a new mobility service that shipped ten internal apps in its first year. The best portfolios leave room for the bet nobody asked for yet.

The operating model

What running a portfolio actually looks like

01

One funnel, one ranked list

All work enters through a single intake and is ranked by impact against leadership's strategic priorities. Nothing gets resourced off-list.

02

Refusing work is a feature

The fastest way to fund what matters is to stop what doesn't: When the analytics team I ran dropped its low-value work, other teams stopped wasting time on it too.

03

Consolidate the overlaps

Merged overlapping solutions, improved workflows, and negotiated favorable enterprise licensing: Seven figures saved annually.

04

Status leaders don't dig for

Progress reporting built for decisions: Time-to-prioritization down 20%, and leadership attention on the objectives that count.

05

The review hour

The weekly portfolio review is where leadership attention gets allocated. Program Sensing, my LLM evaluation, asks whether AI can make those calls with expert judgment.

Also building

[ SHIPPED · LIVE ]
Independent · AI products

Production AI products, end to end

I design, build, and operate LLM products solo: Auth, payments, retrieval, streaming generation, prompt guardrails, analytics, and monitoring. Shipped, live, and run like a portfolio, with data-driven bets and honest kill decisions.

[ PROGRAM SENSING ]
Research · In the field

Program Sensing

An original evaluation of frontier models on portfolio judgment, measured against an expert human baseline. Now recruiting experienced program leaders.

Get in touch

Let's build something
that matters.

Hiring for a senior program role, or exploring LLM evaluation? Here are the ways I can help