What I shipped, where, and what changed.

Today I own a portfolio of enterprise products at McKinsey — an internal app and asset marketplace, GenAI knowledge assistants, a GenAI support agent, and a service-health dashboard — across two engineering squads and an offshore vendor team. I shipped McKinsey's first production GenAI support agent; it worked well enough that other flagship internal teams asked for it, and it spread to five-plus channels in months. I set the 12–18 month roadmap for the marketplace against an ROI model measured in thousands of hours a year, grew active integrations ~30% year over year into the five figures, and scaled a knowledge assistant to hundreds of weekly users at ~80% automated resolution. I also built the answer-quality evaluation the org didn't have — turning “is the AI any good?” from an argument into a tracked signal leadership steers by.

SHIPPED · McKinsey’s first production GenAI agents · LEARNED · one PM + an AI operating model = director-level surface area

The most-used thing I’ve built isn’t at work — it’s on a Mac Mini in my home office. Brain is a small, self-hosted memory server: hybrid search over my Obsidian vault and 2,400+ OCR’d documents, exposed through MCP so every AI tool I use reads from the same memory. No agent framework, no $400/month vector DB. Just hybrid search done right, for $0 a month. It’s the clearest proof of how I think about AI tooling: own your context, keep it boring, make it useful every single day.

SHIPPED · a memory my AI tools read daily · $0/mo · LEARNED · the cheapest edge is context you own

Four years delivering technology for New York City agencies — health, housing, youth services, consumer affairs, emergency management — end to end, from business case to cutover. I wrote proposals that won $25M+ in new work and built the process the firm used to decide which RFPs were worth chasing. Public-sector delivery is unforgiving: fixed budgets, real constituents, no room to hand-wave. It taught me to own the whole lifecycle, not just the fun part.

SHIPPED · $25M+ in won work across five city agencies · LEARNED · own it from business case to cutover

The MBA was where the engineer learned the business half — and I didn’t just study it. I ran the agile process for the business school’s technology center and shipped internal web apps, drove donor analytics for fundraising, got elected to graduate student government (and grew department representation 30%), and TA’d a new-product marketing course. Two national case-competition finishes and Beta Gamma Sigma on the way out.

SHIPPED · internal apps + a student-gov win · LEARNED · product sense is built, not born

I started as an engineer. At Tech Mahindra I led a small team through the kind of work that teaches you what “done” actually means — we held a multi-million-dollar British Telecom account by hitting a hard release with full test coverage and zero live defects. Six people, three cities, one deadline. It’s where I learned that shipping is a discipline, not an event.

SHIPPED · the release that saved the account · LEARNED · done means no surprises in production

Alongside my engineering degree, I ran developer workshops on campus as a Microsoft Student Partner — teaching other students to build on Microsoft’s stack. My first taste of the thing I still do: turning technical capability into something people will actually pick up and use.

SHIPPED · workshops that got students building · LEARNED · adoption is a teaching problem