Yash Agarwal

Machine learning systems — GPU kernels, compilers, quantization. Carnegie MellonIIT KharagpurModularSquarepoint Capital

Hi, I'm Yash, and I love systems and making them run faster. Everything from low level linux internals to gpu stuff. I have fun while engineering code to run on the given hardware, the more heterogeneous the better. My recent work has been on inference engines at Modular, where I worked on everything from FA4 kernels to various new speculative decoding methods to optimizing allocations on the cpu. I'm actively looking for full time roles starting December 2026.

News

  • Joining Deep Learning Systems as a teaching assistant for Fall 2026.

  • Started as an AI Performance Engineering intern at Modular in San Francisco.

  • Started my MS in Intelligent Information Systems at Carnegie Mellon’s School of Computer Science.

  • YesBut, a benchmark for evaluating vision-language models, was published at EMNLP 2024.

  • Joined Squarepoint Capital in Bangalore as a C++ developer.

  • Graduated from IIT Kharagpur with a BTech in Computer Science and Engineering.