Cindy Liu, Product Designer
Headline
7+ years of experience designing for trust where it’s hardest to earn: new paradigms, deep domains, and the wicked problems1 that arise at their intersection.
Footnote
1Wicked problem
A term from design theory for problems with no fixed definition, no finish line, and no right answer, where every attempt has consequences. Coined by Horst Rittel and Melvin Webber in 1973 (opens in a new tab)
Selected work

Defining an MVP for structuring clinical trial data
Turned a strategic bet on nested arm- and outcome-level extraction from clinical trials into a testable MVP, then validated it with trial users.
- Which step of a systematic review could AI actually win?
- How do you turn flat paper-level data into nested arms and outcomes?
- How do you validate a bet before building it?
Designing for a model that’s sometimes wrong
The source viewer, confidence flags, and provenance layer that let biopharma experts verify AI-extracted data, and that survived every product pivot.
- How can experts verify an AI answer in seconds?
- How do you show uncertainty without eroding trust?
- When should the model say “I don’t know”?
Coming Soon

Human-machine systems for defense and health
AI-driven decision support tools for government sectors spanning defense, intelligence, and healthcare.
- When should a watchstander trust a computer-vision detection?
- Which vulnerabilities matter most across an entire enterprise?
- Can AR guide medics through emergency procedures in the field?
Experience
2024–26
Reliant AI (acq. by Cohere)Founding Product Designerfirst designer
2019–24
Charles River AnalyticsHCI Designer I → II44 programs, $43M in funding
2019
MITBachelor of Science, Art & Designself-designed major


