Hey, I’m Allen Lee.
AI safety researcher · generalist engineer · ex-founder in consumer apps.
I’m based in SF/Taipei, working on AI safety, mainly through mechanistic interpretability: trying to understand what safety-trained language models are doing internally, and what safety training and refusal-editing actually change.
If you’re working on making frontier models safer, or building consumer apps that are made with care, I’d love to hear from you 👋