Sevin Varoglu
Senior Software Engineer at NVIDIA
California, United States
Summary
🤩
Rockstar🎓
Top SchoolSevin Varoglu is a Senior Software Engineer with nine years of experience specializing in AI/ML compilers and distributed deep learning systems, now working on the XLA-GPU compiler at NVIDIA. Previously a Principal Engineer at OctoAI, she led teams building model services for generative AI, integrating tooling into LangChain and extending TVM and ONNX workflows for production SaaS. Her open-source contributions to nGraph and TVM include implementing AllReduce/MPI support for distributed training and advancing quantized operator support, directly improving performance on CPU, GPU and accelerator targets. At Intel she drove Habana accelerator enablement, framework integrations, and scalable data-parallel training solutions, alongside lower-level security and graphics work—an uncommon blend of compiler, systems, and security experience. She holds graduate degrees from UC Irvine and combines deep research grounding with hands-on production delivery across cloud and edge ML stacks.
9 years of coding experience
16 years of employment as a software developer
BS Computer Engineering, BS Computer Engineering at Eastern Mediterranean UniversityUniversity of California, Irvine
Github Skills (12)
unit-testing10
compiler10
mpi10
machine-learning10
deeplearning-ai10
c-language10
distributed-training10
deep-learning10
compiler-compiler10
cprogramming-language10
python9
tensor9
Programming languages (3)
Github contributions (5)
NervanaSystems/ngraph
Feb 2018 - Mar 2019
nGraph - open source C++ library, compiler and runtime for Deep Learning
Role in this project:
Back-end Developer
Back-end DeveloperContributions:7 commits, 7 PRs, 31 pushes in 1 year 1 month
Contributions summary:Sevin primarily contributed to the development of the nGraph deep learning compiler and runtime. Their work involved adding and enabling support for the AllReduce operation, crucial for distributed training, including the implementation of MPI support. They also refactored distributed code and added a new "erf" operator, including implementation and unit tests. The user demonstrates expertise in extending the core functionalities of the nGraph library, impacting distributed training capabilities.
inference-enginecppc-librarydeep-learningtvm
apache/tvm
Jan 2022 - Sep 2022
Open deep learning compiler stack for cpu, gpu and specialized accelerators
Role in this project:ML Engineer
ML EngineerContributions:18 reviews, 11 commits, 11 PRs in 7 months
Contributions summary:Sevin primarily contributed to the QNN (Quantized Neural Network) component of the TVM compiler stack. Their work focused on implementing and testing quantized versions of mathematical operations like `rsqrt`, `topk`, `add`, `subtract`, and `multiply`, including per-channel quantization support. The user also modified and tested the `qnn.conv2d_transpose` and `qnn.dequantize` operations, and added support for MeanVarianceNormalization. Their contributions directly impact the performance and functionality of quantized deep learning models within the TVM ecosystem.
metalvulkancompilertensoropencl
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.