Raghu Raja
Principal Engineer at Amazon Web Services (AWS)
Seattle, Washington, United States
Summary
🤩
Rockstar🎓
Top SchoolRaghu Raja is a Principal Engineer based in Seattle with 13 years of experience building high-performance networking, collective communication, and ML-accelerator software and infrastructure. He has led host userspace development for AWS’s Elastic Fabric Adapter and now architects networking and ML accelerator solutions at Annapurna Labs within AWS. Raghu is an active open-source contributor to heavyweight HPC projects like Open MPI and libfabric, where his low-level work on flow control, buffer management, and CUDA buffer handling has improved network performance and reliability. Previously he led ML software ecosystem efforts at Enfabrica and developed extreme-scale userspace filesystems and burst-buffer/datapath features at Cray and LLNL, demonstrating a pattern of solving bandwidth- and latency-critical problems. His background blends deep research (PhD-level study) with production-grade DevOps and systems engineering, often surfacing subtle fixes that eliminate multi-threaded race conditions and optimize real-world cluster behavior.
14 years of coding experience
8 years of employment as a software developer
Doctor of Philosophy (Ph.D.) Computer Science and Engineering, Doctor of Philosophy (Ph.D.) Computer Science and Engineering at The Ohio State University
Github Skills (24)
lib10
fortran10
mpi10
hpc10
openmpi10
c1110
networking10
c1710
low-level-programming10
aws-cloudformation10
aws10
performance-optimization10
network-protocol10
error-handling10
devops10
Programming languages (11)
Github contributions (5)
ofiwg/libfabric
Jun 2018 - Mar 2022
Open Fabric Interfaces
Role in this project:
Back-end Developer
Back-end DeveloperContributions:64 reviews, 263 commits, 239 PRs in 3 years 10 months
Contributions summary:Raghu primarily contributed to the core functionality of the libfabric library, focusing on low-level network interface implementations. Their work involved developing a circular-queue-based sliding receive window utility, addressing naming conflicts in the EFA provider code, and optimizing buffer pool allocation. They also made updates to the EFA provider, including improvements to format specifiers, flow control, and handling of CUDA buffers in the send and receive paths, signifying contributions related to network performance and reliability.
interfacesminecraftfabric
aws/aws-parallelcluster
Aug 2017 - Mar 2018
AWS ParallelCluster is an AWS supported Open Source cluster management tool to deploy and manage HPC clusters in the AWS cloud.
Role in this project:DevOps Engineer
DevOps EngineerContributions:1 release, 10 commits, 15 PRs in 7 months
Contributions summary:Raghu primarily contributed to the AWS ParallelCluster project by addressing build failures, adding new CLI options, updating the CHANGELOG, and improving configuration clarity. The user fixed Python 3 build issues, added a new option to the 'stop' command to handle compute fleet only, removed the support blurb about C5 support, and updated CHANGELOG and configuration documentation. Additionally, the user modified the start command to query CloudFormation for the template and dropped the --reset-desired parameter.
awsaws-cloudhpc-cluster
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.