Rohit Santhanam
Senior Member Of Technical Staff at AMD
Austin, Texas, United States
Summary
🤩
Rockstar🎓
Top SchoolRohit Santhanam is a Senior Member of Technical Staff at AMD in Austin with five years focused on enabling and optimizing machine learning frameworks and compilers for AMD/Xilinx AIE and ROCm GPUs. He combines deep systems and performance engineering experience—tuning XLA and adapting JAX for ROCm, resolving MI200 performance issues and enabling ROCm unit tests—with a strong background in low-level, vectorized C++ code dating back to embedded and SDK work. Rohit’s work bridges compiler internals, hardware-specific solver APIs, and build/DevOps improvements, so he’s as comfortable fixing segfaults in MLIR as streamlining large-scale CI and builds. His career includes hand-optimized image processing on SIMD instruction sets and product-facing performance wins at companies from Thorlabs to Roku, showing a practical focus on shipping robust, high-performance systems.
5 years of coding experience
22 years of employment as a software developer
BSEE, Electrical Engineering, BSEE, Electrical Engineering at Case Western Reserve UniversityMSEE, Electrical and Computer Engineering, MSEE, Electrical and Computer Engineering at University of Pittsburgh
Tamil
Github Skills (18)
c-language10
python10
gpu-programming10
roc10
performance-optimization10
xla10
compiler10
amdgpu10
jax10
cprogramming-language10
linear-algebra10
llvm9
mlr9
build-system9
hip9
Programming languages (5)
Github contributions (5)
openxla/xla
Jul 2021 - Jan 2023
A machine learning compiler for GPUs, CPUs, and ML accelerators
Role in this project:
Back-end Developer & Performance Engineer
Back-end Developer & Performance EngineerContributions:35 commits in 1 year 5 months
Contributions summary:Rohit's contributions primarily involve optimizing the XLA compiler for GPU execution, with a particular focus on AMDGPU (ROCm) and improving performance on the ResNet50 model. Their work includes addressing seg fault issues, adapting to upstream changes impacting ROCm performance, and fixing MI200-related performance problems. The user also worked on the BEF thunk for ROCm, addressing MLIR-related GEMM tests, and enabling unit tests for ROCm. Their work also includes modifying the RoundNearestEven operation.
compilercommunity-drivenmachine-learningmodular
jax-ml/jax
Mar 2022 - Oct 2022
Composable transformations of Python+NumPy programs: differentiate, vectorize, JIT to GPU/TPU, and more
Role in this project:Back-end Developer & DevOps Engineer
Back-end Developer & DevOps EngineerContributions:21 reviews, 19 commits, 20 PRs in 7 months
Contributions summary:Rohit primarily focused on integrating and adapting the JAX library for ROCm (AMD GPU) support. Their contributions included consolidating and adjusting various linear algebra APIs (e.g., cuSolver, hipSolver), fixing ROCm-specific compilation issues, and enabling unit tests for the ROCm platform. They also updated the build configuration and dependencies to support ROCm and enhanced the build process.
pytorchpythonjitautomatic-differentiationgpu
Find and Hire Top DevelopersWe’ve analyzed the programming source code of over 60 million software developers on GitHub and scored them by 50,000 skills. Sign-up on Prog,AI to search for software developers.