Utkrisht Patesaria

About Utkrisht Patesaria

Utkrisht is a member of AMD’s pre-silicon team for MI SoC, where he focuses on developing performance models and optimizing AI workloads for the next generation of data-center GPUs. His work expands from writing GPU kernels using Triton/ROCm, going down the stack to assembly-level, extracting the architectural performance of AMD GPUs.He recently earned his graduate degree from the Indian Institute of Science, Bangalore, under the supervision of Prof. Arkaprava Basu in the Computer Systems Lab. He primarily worked on the CPU-GPU memory subsystem interface, in turn optimizing the GNN training pipeline. Previously, he worked at Nutanix as an SRE, troubleshooting issues across the entire cloud stack.

Experience

1. Sr Silicon Design Engineer

   AMD

   Aug 2025 — Present · Bengaluru, IN

   Part of AMD’s Data Center Group, focused on pre-silicon performance modeling for next‑generation MI (Instinct) GPU SoCs. Roles and responsibilities are broadly divided into- Optimizing GEMM/Attention and collective kernels in Triton/ROCm to harness the power of upcoming GPU generations by exploiting architectural enhancements. Closely working with IRIS(https://github.com/ROCm/iris)- Dissecting the GPU pipeline to identify bottlenecks for various kernels—primarily GEMM/Attention variants—and proposing architectural-level decisions (e.g, cache modeling) to improve performance against target metrics such as bandwidth, FLOPs, and latency- Producing performance projections that map kernel-level behavior to workload-level estimates.

Education

• Indian Institute of Science (IISc)

  Mtech Research , Computer Science Automation

• St. patricks school

  10th

  2003 — 2014

• Kalinga Institute of Industrial Technology, Bhubaneswar

  Bachelor's degree, Computer Software Engineering

  2016 — 2020

• Hemsheela Model School

  Intermediate, PCM

  2014 — 2016