I am a Staff AI Research Scientist at Meta Superintelligence Labs. In my past life I was a professor of theoretical condensed matter physics at Brown & UMD.

My current research interests are 

• Science of LLMs: init, scaling laws, post-training, evaluation, interpretability, sampling.

• Science of data: toy models, artificial data, model collapse, curricula.

• Science of reasoning: inference/pre-training trade-offs, RL for LLMs.

• Science of efficiency: pruning, quantization, distillation, efficient architectures.

• AI for theoretical sciences.

• Emergent behavior in complex systems.

Some AI topics I worked on: initialization, grokking, interpretability, pruning & quantization, model collapse, neural scaling laws, self-organizing modular architectures.

Some physics topics I worked on: quantum hall effect, topological phases of matter, fractons, classical and quantum liquid crystals, cold atomic gases, chiral active matter, integrable systems.

See more on Google Scholar!