Loading...
Loading...
Training open-source LLMs to detect synthetic text via watermark learning.
200k CHF deployed, 176 startups evaluated, and meeting some of the coolest people I know.
2nd of 152 teams at Spaceport America Cup, 1st at EuRoC. Made lots of things blow up (while testing of course).
PM on financial services at one of Europe's top scaleups. Saw firsthand how high-performing teams ship.
Simulation-to-chip pipeline for neuromorphic computing. Saw how production codebases are built.
Memento vivere
TEXT TBC. More emphasis on "building" and genuine technical things. from how embedding space topography can affect LLM watermark effectiveness,
to what a tea leaf is biochemically doing in the hours between plucking and becoming tea.
I tend to pick up interests and then go unreasonably deep:
building geography olympiad medals from age 10,
12 years of clarinet and piano across concert halls and living rooms.
Still learning every day, and always will be :)
A few of those rabbit holes, below.
Hover to explore