Mithun Radhakrishnan
Principal Sys SW Engineer at NVIDIA
San Jose, California, United States
Summary
🤩
RockstarMithun Radhakrishnan is a Principal Systems Software Engineer at NVIDIA with nearly two decades of experience building and optimizing large-scale data and firmware systems. He specializes in accelerating Big Data workloads on GPUs and has contributed backend improvements and JNI bindings to high-profile open-source projects like RAPIDS cuDF and the Spark RAPIDS plugin. Previously he spent a decade at Yahoo scaling Apache Hive and HCatalog to internet-scale workloads, focusing on metastore performance and queries across hundreds of thousands of partitions. His early career writing firmware at Hewlett-Packard gave him deep systems-level debugging and self-healing insight, which informs his work on performance-critical distributed systems. Based in San Jose, he blends low-level C++ and firmware expertise with distributed data engineering, and is known for pragmatic optimizations—particularly around windowed aggregations and tricky floating-point corner cases. Outside of work, colleagues note a wry sense of humor about C++ withdrawal that belies a fierce attention to detail.
10 years of coding experience
18 years of employment as a software developer
English
Github Skills (23)
apache-spark10
c-language10
spark10
gpgpu10
pytest10
python10
testing10
dataframes10
big-data10
cudf10
data-structure10
dataframe10
rapids10
window-functions10
gpu10
Programming languages (7)
Github contributions (5)
NVIDIA/cudf
Mar 2020 - Jan 2023
cuDF - GPU DataFrame Library
Role in this project:
Back-end Developer
Back-end DeveloperContributions:1787 reviews, 154 commits, 118 PRs in 2 years 10 months
Contributions summary:Mithun appears to be focused on enhancing the documentation and performance of the cuDF GPU DataFrame Library, particularly around window-based aggregations. The commits demonstrate a focus on optimizing existing rolling window functionalities by parameterizing the null comparator behavior within join operations, as well as making changes to accommodate more use cases. These changes include implementing optimizations to the `COLLECT` and `NTH_ELEMENT` window aggregations and adapting window functions to accommodate non-fixed-width types and handle edge cases within range queries. The user's code modifications include contributions to the JNI bindings for the library.
cudfdataframegpurapidsarrow
NVIDIA/cudf-spark
Jun 2020 - Jan 2023
NVIDIA cuDF for Apache Spark plugin - accelerate Apache Spark with GPUs
Role in this project:Back-end Developer / Test Automation Engineer
Back-end Developer / Test Automation EngineerContributions:519 reviews, 43 commits, 196 PRs in 2 years 7 months
Contributions summary:Mithun primarily worked on developing and implementing integration tests for the `spark-rapids` plugin, specifically focusing on handling corner cases related to NaN/zero values in floating-point operations and window functions. They contributed to test cases involving hash aggregates and distinct aggregates. Their work also included refactoring the existing test code and introducing a utility for comparing SQL query results between CPU and GPU.
apache-sparkcudfnvidiasparkgpu
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.