04Cosmetic safety
SkipLab


- Role
- Demo code, ML workflow and technical pitch
- Team
- Five UC Berkeley students
- When
- January 2025 to the stage on June 14, 2025
- Made
- A stage prototype for animal-free safety screening
Status Won 1st Place, $7,000. Stage demo with illustrative values.
Context
In January 2025, five of us at Berkeley started with one question: does a new skincare formula really need to be tested on an animal first?
SkipLab was our answer: a screening step that would read a formula before it reaches the lab, flag risky ingredient combinations, and point to the non-animal tests worth running. It was designed to learn from public, peer-reviewed skin-safety data.
What we built
We made the idea something a judge could click through: a prototype that goes from uploading a formula to an analysis, a safety score and a regulatory report.
I wrote the demo code and designed the ML workflow, using peer-reviewed skin-sensitization and toxicology datasets.
On stage
We pitched in June 2025 at the Organoid Developer Conference in Seoul. In a lab coat, I presented the technical half with a teammate, while a toy rabbit named Brian stood in for every animal the tool might spare.
The numbers on screen were illustrative and no model ran behind them, but a judge could follow every step.

Outcome
1st Place, $7,000, in the inaugural AFI Global Idea Challenge, Seoul, June 14, 2025.
SourceAFI prize email, June 26, 2025; team certificates, July 22, 2025
UC Berkeley’s SCET told the story of the win in a feature about me.
SourceSCET article, September 4, 2025
Proposed, not yet done
A screening model trained on public, peer-reviewed skin-safety data. It was designed for the pitch and never trained.
Links
- SCET article on the winSeptember 2025
Credits
- Pitched
- June 14, 2025
- Photos
- Animal Free Initiative