I design the systems underneath data-heavy software.

1. 

   Aspira · Rainger, one front desk for public lands

   The software agencies use to run their parks had grown by acquisition into seventeen applications. Collapsed them onto one bookable object with a unit of time. Routine reservation 64% faster.

   • Platform
   • Govtech
   • Domain model

   2024 — 26 Case study →

2. 

   Lily AI · Catalog manager for retail attribution

   Re-anchored a 40M-SKU catalog on delivery batches instead of one flat list. Merchandisers stopped searching a graveyard of discontinued product. Time to locate a batch fell 71%.

   • Information architecture
   • Enterprise SaaS
   • Data tooling

   2022 — 24 Case study →

3. 

   Torch · Learning goals for executive coaching

   Clients paid for coaching and ignored the software. Replaced percent-complete goals with Growth Points derived from 360 survey data. Weekly active use rose 3.4× in two quarters.

   • 0 → 1
   • Data model
   • B2B SaaS

   2019 — 21 Case study →

4. 

   LPL Financial · Practice Hub

   Advisors could see their numbers but not whether the numbers were good. Four metrics, peer percentile benchmarks, and preset S.M.A.R.T. goals. 62% of pilot advisors set a goal in week one.

   • Fintech
   • Benchmarking
   • Dashboard

   2021 — 22 Case study →

5. 

   GoodUnited · Reporting in plain English

   Replaced a six-spreadsheet monthly reporting ritual with one paragraph a director could forward without editing. Report assembly went from roughly four hours to under two minutes.

   • Analytics
   • Natural language
   • Nonprofit

   2020 — 21 Case study →

6. 

   XPO Logistics · Consumer delivery tracking

   Heavy freight needs someone home. Gave Peloton customers a real delivery window and a way to change it, instead of "out for delivery." Where-is-my-order calls fell 34%.

   • Logistics
   • Mobile web
   • Status design

   2017 — 18 Case study →

7. 

   Barkd · A marketplace built on distrust

   Puppy buying happens offline, in cash, with no recourse. Escrowed deposits, verified breeders, in-app messaging, and a lifestyle match score. Shipped to both stores at 4.6 stars; seed round closed.

   • Marketplace
   • iOS + Android
   • 0 → 1

   2019 — 20 Case study →

8. 

   Healthcare scheduling · Booking as a checkout

   A patient said she was "shopping for a doctor," so I built a checkout instead of a form. HIPAA-safe intake, running order summary, calendar sync. Booking completion hit 71% in usability testing.

   • Healthcare
   • HIPAA
   • Conversion

   2018 Case study →

9. 

   StrikeFeed · Designed, built, launched in 7 days

   Self-initiated. I had mentored 5,000 designers who had never seen a product go all the way out the door, so I shipped one in public in a week — brand, UI, CMS, server, launch.

   • Self-initiated
   • Full stack
   • Brand

   2018 Case study →

How I work

Model the domain first. Interfaces are the last thing I draw.

Anyone can generate a screen now. The expensive part is knowing which screen should exist — and what the things on it are actually called.

• Definition

  The fastest way to fix a screen is usually to fix the noun underneath it. I spend the first weeks arguing about what things are called, until the product speaks the user's vocabulary instead of the database's.

• Systems

  Borrow the mental model before inventing one. A clerk working a queue of 200 permit applications is doing the same job as anyone triaging an inbox — select, act, move on. Apple Mail solved that in 2003. I ship migration tooling and office hours, not mandates.

• Craft

  Dense expert software rewards precision: honest empty states, legible tradeoffs, numbers you can trace back to how they were measured, and no motion that hasn't earned its frame.