about
blog
now
publications
projects
cv
more
repositories archive
contact
licenses privacy
docs: UPIR docs: AI Metacognition

Interpretability

an archive of posts in this category

Dec 20, 2025	I Trained Probes to Catch AI Models Sandbagging

Subhadip Mitra

Technical Leader, Inventor, and Researcher building the future of Data & Applied AI.
Leading Google Cloud's D&A practice across Southeast Asia.

contact@subhadipmitra.com

Explore

About
What I'm Doing Now
Blog
Publications
Repositories

Connect

Email
Schedule a Call
LinkedIn
GitHub

© 2025 Subhadip Mitra. Some Rights Reserved. Last updated: December 20, 2025. Sitemap · RSS · LLMs