Lightning talk at Frontiers of Online Reinforcement Learning Workshop, IMSI
Talk title: "An Oracle Taxonomy for LM Post Training".
PhD Student in Computer Science, CU Boulder
Hi! I am Amit Kiran Rege, a PhD student in Computer Science at the University of Colorado Boulder advised by Claire Monteleoni.
My research is in theoretical machine learning broadly construed, especially problems in AI safety and control. My recent focus is applying reinforcement learning theory to LLM post-training, while maintaining interests in foundational RL theory and the foundations of interpretability.
I am also interested in startups and early-stage venture through my role as an Associate at the Deming Center Venture Fund, where I help source and evaluate early-stage Colorado startups.
Outside work, I enjoy soccer, cricket, music, and following current affairs.
Quick highlights for talks, releases, paper milestones, and travel.
Talk title: "An Oracle Taxonomy for LM Post Training".
Thesis title: "Structure and Reliability in Interactive Decision Making".
Presented our ongoing work on LLM reasoning.
Featured papers with links to PDFs, code, slides, and project pages.
Accepted at the Twenty-Ninth International Conference on Artificial Intelligence and Statistics.
Accepted oral presentation at the 8th Annual Learning for Dynamics and Control Conference.
Accepted at the 8th Annual Learning for Dynamics and Control Conference.
Long-form technical notes and research explainers.
Under Construction
Coming soon!
Interactive prototypes and research engineering experiments.
Static retrieval debugger that shows where RAG failed: chunk split, metadata filter, or reranker regression. Each case includes gold support spans, counterfactual fixes, and answer claim audits.
Keep CV/Resume files in `public/assets/docs/` and update links in `src/data/profile.json`.