Anshuman Agrawal #

I am a researcher and engineer interested in the efficiency of intelligence.

My work lies at the intersection of High-Performance Computing (HPC) and Machine Learning. I am currently investigating how to make large-scale foundation models accessible on edge hardware through quantization, sparsity, and distillation.

I believe that true understanding comes from building systems from scratch. This site serves as a digital garden—a collection of technical notes, research logs, and essays on the philosophy of mind and artificial cognition.

Current Focus #

• Systems for AI: Benchmarking Transformer inference on NVIDIA H100s.
• Efficient ML: Post-training quantization (INT8) and Knowledge Distillation.
• Deep RL: Stochastic processes in continuous control environments.

I don’t just run code; I try to understand the why behind the how.