Case Records
// CASE RECORDS
A case record is how I run work: one problem, one owner, staged from build through adoption to scale. Every case names its measures up front and reports against them honestly — a target stays labeled a target until it has been measured. Most of what follows is AI and automation put to actual use in a small business — mine, and the ones I’ve worked for — with the people kept exactly where their judgment matters and the result measured in the business’s own units. Three registers below: the work delivered for clients and employers, the delivery systems I run in production, and the strongest entries from my studio’s record — a quarter century of dated, checkable work.
Delivered, and measured
The records behind the four lines I take directly. Each carries a result the business already tracked, measured before and after.
Incident Response for a $630M Live-Commerce Program — program technical lead for incident response on the FIFA World Cup 2026 premium-hospitality program at TKO: a 70-person operation that took over $630M in cash without a notable loss, key page-load times cut by more than six times, readiness run from a written checklist. The record behind incident readiness.
Event Apps That Lifted Revenue 12–17% — lead developer on event mobile apps for On Location / Endeavor: identity resolution and customer metrics that reached six million users in a first week and lifted event revenue by 12% and 17%, attendance by 15%. The record behind measurement.
A 30-Day Analyst Process Cut to Five Seconds — at Goodwater Capital, an AI-driven service replaced a thirty-day manual process with a five-second task and three days of human review, and twenty-plus pipelines went from a day’s processing to hours on a tenth of the machines. The record behind AI workflow readiness.
Marketing Technology at Trillion-Transaction Scale — at Signal, an acquired email-marketing platform integrated into the core product, first-party data segmentation shipped, and tooling for seventy-plus engineers held at six-sigma uptime. The record behind measurement and fractional CTO.
On-Demand ML Infrastructure for a Research Lab — standing technical lead, consulting, for a Caltech bioengineering group: GPU resources used on demand instead of by reservation, reproducibility built into the pipeline, a development practice the team didn’t have before. What a retainer looks like.
Systems in production
- Self-Driving Agentic SDLC Factory — my own delivery loop, now scaling: a prioritized issue backlog goes in; reviewed, squash-merged software comes out, with a person only where a person matters.
- On-Call Schedule Shift Automation — a recurring manual chore, encoded once and run automatically inside the team’s observability tooling. In adoption.
- AI-Augmented SDLC for Legacy Migration — a governed, AI-augmented pipeline for migrating thousands of legacy stored procedures, targeting 30–50% lower cycle time at comparable or better quality.
- Always-On Apartment-Search Lead Engine — Find a Pad, my own system: every listing source watched, every find scored and scam-screened, the phone buzzing only when a lead is worth acting on.
- Reliability OS Platform Migration — Roger, my own consistency system, mid-migration from a single macOS authority to a Firebase-native platform with Firestore as the one source of truth.
- Low-Ticket Consumer Clarity Product — Read The Room, my own consumer product: a personalized clarity report delivered on a serverless stack where the unit economics are the design problem.
- Governed Influence-Growth Operator — a platform-agnostic growth system in which every outbound action waits for a human confirm: scale without spam.
From the studio record
The entertainment record is a different book, and it lives with my studio — dated, checkable, back to 2001. These five carry the full write-up — goal, constraints, force, time, result — on their pages:
David Blaine — Drowned Alive — I built and ran the full digital presence for a live seven-day stunt at Lincoln Center: davidblaine.com held national traffic at over a million unique visitors a day, ~10-minute average visits, 300,000 fan messages collected. The engagement ran about six years, through the ABC-specials era.
U.S. Department of Justice — “Know Where They Go” — I co-directed a :30 internet-safety PSA for a federal client, delivered to network clearance and aired nationally on NBC, CBS, Fox, and Oxygen.
Rubblebucket — two music videos — location shoots, live choreography with Juilliard dancers, delivered in sequence across two albums; Down in the Yards sits at six-figure views on the band’s official channel.
The XJ music patent — I took the adaptive-score engine’s core architecture through the full USPTO process as a solo inventor: granted 2019, assigned to my studio, in force and licensable through 2038.
Fever Tilt — the current build, shipping in public: an original 1v1 arena sport, solo-built, proven through weekly live play at Noisebridge, shown in the Ribbon-winning Coffee Table Arcade exhibit at Maker Faire Bay Area 2025, Steam Early Access this November.