Hi, I'm Michael 👋

I help teams make
better decisions,
not just better screens.

I work at the intersection of product strategy, AI, and marketplace complexity. My job is to turn high-stakes, ambiguous problems into clear decisions — then design the experiences that make those decisions stick.

14+ Years in product design
$2B+ Marketplace GMV influenced
4 Core product areas led
Based in Austin, Texas
About

I've always worked upstream.
AI gave me more time to stay there.

I'm a Principal Product Designer at ACV Auctions. I design how cars get bought and sold at ACV Auctions — from first touch to post-auction — turning messy, fragmented data into clear decisions through better disclosures and AI-powered condition reporting.

A prototype that used to take two weeks now takes two hours. That's not a reason to step back from production — it's what makes both possible. I use AI to move fast in Figma and in code, which means I can generate real surface area to test ideas without losing the time I've always spent upstream: questioning the brief, building alignment, defining what success looks like before any pixel moves.

Problem framing has always been part of the job for me — not something I shifted into. The work I'm most proud of lives in two places: the decisions made before anyone opened Figma, and the shipped product that made those decisions real. The throughline: complex systems, real constraints, decisions that cost something.

Let's talk →
How I operate
01
I work upstream The best design leverage is in how a problem gets scoped, not how it gets solved. I push to be in the room before the brief exists.
02
Alignment is the deliverable A perfect design no one believes in ships as a bad product. I spend as much time building shared understanding as building interfaces.
03
Show the trade-offs Every direction you choose means rejecting others. I document what I said no to — and why — so decisions don't have to be relitigated six months later.
04
Use AI to buy back strategic time Production is no longer the bottleneck. I use AI to rapidly prototype, generate variants, and compress execution cycles — so the hours I protect are spent on problem framing, not pixel-pushing. The leverage is upstream.
Selected Work

Four problems.
Four sets of hard decisions.

2024
Trust Marketplace AI

Vehicle Detail Page Redesign

Dealers make $20,000+ decisions in under 4 minutes on a mobile device. The old VDP gave them everything except what they actually needed to commit.

Key decision We had to choose between showing more data (transparency) and showing less, better-organized data (decision speed). We chose the latter — and had to defend it internally for 6 weeks.
+22%
Buyer bid conversion
−34%
Post-sale arbitration
−2.1 min
Avg. time to bid decision
2023
Search Scale Filtering

Search & Results Redesign

100,000+ active listings. Dealers with 5 minutes between lot walks. A filter system built for desktop power users who were now 80% on mobile.

Key decision We removed 14 filters from the default view. Every PM in the room hated it. Six weeks post-launch, search-to-bid conversion was up 18% and support tickets about "couldn't find it" dropped by half.
+18%
Search-to-bid conversion
−47%
Abandoned search sessions
2023–24
AI Trust Condition

AI Condition Reports

ACV's computer vision model could generate condition reports faster and cheaper than human inspectors. The model was ready. The UX problem was trust — and we got it wrong the first time.

Key decision Version 1 showed AI confidence scores on every field. It felt transparent. In user testing, it made dealers less confident, not more. We rebuilt the trust model from scratch around outcomes, not methodology.
4.1→4.7
Report trust score (1–5)
−34%
Post-sale disputes (AI reports)
2022
Design System Platform Ops

Inspector Operations Platform

600+ field inspectors. 3 legacy tools. A ground-up redesign that had to work on Android in a parking lot in January with one hand, while the other held a flashlight.

Key decision Structured input or free-form? We shipped both — and let usage data pick the winner. Expert inspectors adopted structured mode at 91%. New inspectors never left guided mode. Both metrics went up.
−28%
Data error rate
+9 min
Saved per inspection
91%
Expert mode adoption
Let's Talk

Looking for a designer who can lead and still do the work?

I'm open to Staff and Principal IC roles at companies building complex, high-stakes products. If that's you — let's talk.