Sudhir Kylasa
Senior Member Of Technical Staff at AMD
Milpitas, California, United States
Summary
🤩
Rockstar🎓
Top SchoolSudhir Kylasa is a Senior Member of Technical Staff and PhD-trained engineer with over a decade of experience bridging academic research and production systems in high-performance computing, distributed graph processing, and telecommunications. He has led global engineering teams at Alcatel-Lucent and built GPU-accelerated solvers and multi-GPU training pipelines during his PhD and postdoc at Purdue, work that produced widely used tooling and publications. More recently he scaled graph preprocessing for trillion-edge datasets at Amazon and optimized multi-head attention for LLMs on AMD MI-series GPUs for top-tier clients like Meta and Microsoft Research. A pragmatic researcher-developer, Sudhir contributes to open-source graph deep learning infrastructure (dgl) by hardening distributed training and memory-efficient partitioning. Based in Milpitas, he blends systems-level expertise (CUDA, distributed systems, DevOps) with telecom-grade production experience, making him fluent at turning research into deployable, high-performance solutions.
8 years of coding experience
21 years of employment as a software developer
MS Computer Networks, MS Computer Networks at The University of Texas at DallasDoctor of Philosophy (Ph.D.) Computer Engineering, Doctor of Philosophy (Ph.D.) Computer Engineering at Purdue University
English, Hindi, Telugu
Github Skills (16)
distributed-systems10
graph-neural-network10
python10
dgl10
data-pipelines9
numpy9
fileio9
file-access9
data-pipeline9
deep-learning9
file-handling9
memory-optimization9
file-processing9
pytorch8
machine-learning7
Programming languages (3)
Github contributions (5)
dmlc/dgl
May 2022 - Nov 2022
Python package built to ease deep learning on graph, on top of existing DL frameworks.
Role in this project:
Back-end Developer & DevOps Engineer
Back-end Developer & DevOps EngineerContributions:394 reviews, 18 commits, 96 PRs in 6 months
Contributions summary:Sudhir implemented critical components of a distributed training pipeline for deep learning on graphs. They focused on data processing, including reading graph files, shuffling data, and generating partition-specific files. Their work also included debugging and fixing issues within the distributed training infrastructure, notably in the areas of message passing and feature handling. Furthermore, they made significant contributions to optimizing memory usage and enhancing the robustness of the distributed graph partitioning pipeline.
pytorchpythondeep-learningmachine-learninggraph-neural-networks
kylasa/dgl
Apr 2022 - May 2023
Python package built to ease deep learning on graph, on top of existing DL frameworks.
Contributions:1 PR, 246 pushes, 56 branches in 1 year 1 month
pytorchpythondeep-learningmachine-learninggraph
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.