Uber
Results
Users impacted
5M+
U.S. earners covered by a redesigned tax form
Documents generated
2M+
Tax forms generated
Compliance
5.6% -> 23%
Completion moved from 5.6% toward a compliant threshold
Team

Product Design



Product Management & Engineering
Tech stack



Design & AI
Role
Led end-to-end design and was embedded with the research team from the first redesigned iteration through public launch. Owned the product experience in full — auto-fill logic, inline validation, and the review flow — working directly with compliance and engineering along the way.
Overview
Tax profile completion sat at 5.6% among U.S. drivers and couriers, despite being mandatory for IRS 1099 reporting. Uber already had the data it needed — legal name, SSN, address — but only SSN carried into the form, and validation ran offline after submission with no way to catch a mismatch before it became a support ticket or a $180 penalty Uber had already absorbed once. I redesigned the flow to auto-fill known data, validate inline, and prompt review proactively — turning a legal obligation nobody understood into a guided path earners could complete correctly in one sitting.
Insights
One platform. 3 user worlds
Funds, portfolio companies, and corporates each had fundamentally different goals. The experience had to adapt to all three without fracturing into three separate products.
Investors wanted the hard numbers
Surface-level summaries weren’t enough. Real-time financials and cap table data needed to be embedded directly in profiles.
Pitch fatigue was real
Investors were overwhelmed by irrelevant deals, demanding high-precision filtering by niche, stage, and sector.
Trust over trendiness
In high-stakes finance, the interface had to feel authoritative, transparent, and credible — not just visually appealing.
Explorations & iterations
Before committing to a direction, I explored three distinct concepts — stress-testing the balance between legal compliance and user speed through structured A/B/C testing with real earners.
Work

AI demo
I used AI-assisted tooling to prototype the actual interactions — auto-fill, inline validation, and the review flow — so you can click through it instead of judging it from static screens alone. It's a prototype, not production code, but the logic mirrors the real design decisions.













