Bhavani Subramanian
Deep Learning Software Engineer at Intel Corporation
Hillsboro, Oregon, United States
Summary
👤
Senior🎓
Top SchoolBhavani Subramanian is a Deep Learning Software Engineer with eight years of experience specializing in C/C++ development, parallel computing, and performance optimization for ML frameworks. Based in Hillsboro, Oregon, she has driven low-level, architecture-aware improvements at Intel, optimizing oneDNN and integrating oneDNN v3.x into TensorFlow to accelerate convolutions, pooling, normalization and fused kernels. Her work spans backend kernel tuning (including AVX2-optimized dilated convolutions and backward-data paths), primitive cache fixes, and microarchitectural bottleneck analysis for data-center workloads. She blends systems-level performance engineering with practical ML engineering, moving optimizations from prototype into production-grade libraries used widely across the ML ecosystem. Former PCB routing experience and an engineering background from Visvesvaraya Technological University underpin a pragmatic, hardware-conscious approach to software performance. Actively contributing to high-profile open-source projects, she seeks full-time or internship roles where deep performance expertise and low-level optimization deliver measurable speedups.
8 years of coding experience
5 years of employment as a software developer
Bachelor’s Degree, Electrical, Electronics and Communications Engineering, Bachelor’s Degree, Electrical, Electronics and Communications Engineering at Visvesvaraya Technological UniversityMaster’s Degree, Computer Science, Master’s Degree, Computer Science at Georgia Institute of Technology
English
Github Skills (21)
bfd10
c-language10
performance-monitor10
edn10
avx10
machine-learning10
performance-analysis10
deep-learning10
tensorflow10
deep-neural-networks10
neural-network10
convolution10
cprogramming-language10
normalize9
normalization9
Programming languages (2)
Github contributions (5)
uxlfoundation/oneDNN
Oct 2019 - Nov 2021
oneAPI Deep Neural Network Library (oneDNN)
Role in this project:
Back-end Developer & Performance Engineer
Back-end Developer & Performance EngineerContributions:72 commits, 1 comment in 2 years 2 months
Contributions summary:Bhavani primarily contributed to optimizing the oneAPI Deep Neural Network Library (oneDNN) for improved performance. Their work involved enabling and optimizing dilated convolutions for the AVX2 instruction set architecture (ISA), specifically focusing on the backward data path. Furthermore, they fixed a typo and made improvements related to pooling and batch normalization kernels. Their contributions demonstrate a focus on low-level optimization techniques within a deep learning library.
bfloat16sse41avx512openmpopencl
tensorflow/tensorflow
Dec 2017 - Aug 2022
An Open Source Machine Learning Framework for Everyone
Role in this project:ML Engineer
ML EngineerContributions:46 reviews, 67 commits, 38 PRs in 4 years 8 months
Contributions summary:Bhavani's commits primarily focus on adding support for oneDNN v3.x within the TensorFlow codebase, specifically related to various machine learning operations. This includes integrating oneDNN for convolution, maxpool, avgpool, softmax, layer normalization, and fused operations like fused-mish and fused batchnorm. The user's contributions involve adapting existing kernels, adding new functionalities, and optimizing performance by utilizing the oneDNN library's capabilities. Additionally, the user is involved in fixing primitive caching issues, demonstrating a deep understanding of optimization within the context of this machine learning framework.
pythondata-sciencedeep-learningmlmachine-learning
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.