FAQ

Who is this?

Dhairyashil R. G., a performance engineer in Bangalore. Senior Member of Technical Staff at AMD on the uProf profiler; PhD from the Indian Institute of Science. He measures where time, cost and failures go in compilers, parallel code and AI agents, and publishes the measurement so it can be re-run.

What is the current role?

Senior Member of Technical Staff, AMD, profiling team (uProf), since April 2024. Before that, tech lead on a C++/MPI particle simulator at SankhyaSutra Labs, 2018–2024.

What are the three strongest proofs?

1. Upstream pull requests to LLVM/MLIR and Spack, 8 pull requests, 5 mergedsourcesrc/data/prs.json (tools/fetch_prs.py)checked2026-09-23kindmeasured; the live states are on Work.
2. A pass that removes every masked load from a PolyBench kernel on AArch64, 14 → 0sourcehttps://github.com/llvm/llvm-project/pull/215340checked2026-09-22hardwareAArch64 NEON, Apple silicon (no SVE); not measured on SVEcommandbench/run.shkindmeasured, taking it from 14.70 → 2.89 mssourcehttps://github.com/llvm/llvm-project/pull/215340checked2026-09-22hardwareAArch64 NEON, Apple silicon (no SVE)commandbench/run.shkindmeasured wall clock. Reproduce it.
3. A DSMC simulator 2×sourceSankhyaSutra Labs internal benchmark (2018–2024), résuméchecked2026-09-21kindself-reported faster than Sandia’s SPARTA on the same hardware, scaled past 100sourcerésumé, SankhyaSutra Labschecked2026-09-21kindself-reported nodes (self-reported; internal benchmark).

How do I reproduce a number?

Every number on the site has a receipt (hover or tap it): a source, a date checked and, where one exists, a command. The commands for the compiler results are in mlir-inbounds-harness; the agent studies’ run ledgers are in agent-harnesses-2026.

What is he open to?

ML compiler and inference-performance work (MLIR, LLVM, kernel and serving performance), and agent infrastructure and evaluation. Remote or Bangalore.

How to contact?

hello@dhairyashilrg.dev. Profiles: GitHub, LinkedIn, YouTube, ORCID, Google Scholar.

How should I cite this site?

Dhairyashil R. G., dhairyashilrg.dev, the page, and the date printed on it. Machine-readable copies: /llms.txt, /cv.json.