Kanvi Khanna
Software Engineer at Intel Corporation
San Diego, California, United States
Summary
🤩
Rockstar🎓
Top SchoolKanvi Khanna is a software engineer with seven years of experience building and optimizing machine learning systems, currently based in San Diego and working at Intel. She has contributed performance-focused improvements to high-profile open-source projects like TensorFlow, implementing a new fusion pattern for _MklFusedBatchMatMulV2 that improves runtime efficiency. At Intel she’s applied model-level optimizations for medical image segmentation (3D-Unet on BraTS), adding int8 and bfloat16 support and tuning both accuracy and performance pipelines for deployment. Her background combines rigorous academic training—an MS in Computer Science from San Diego State University—with hands-on engineering that spans graph-level optimizations and production readiness. Colleagues describe her as detail-oriented and pragmatic, able to bridge research-quality models and real-world performance constraints. She brings a quiet focus on measurable impact: faster inference, tighter precision control, and robust test coverage.
7 years of coding experience
Master's degree, Computer Applications, 76%, Master's degree, Computer Applications, 76% at GNDUMaster of Science (MS), Computer Science - part time, 3.8/4.0, Master of Science (MS), Computer Science - part time, 3.8/4.0 at San Diego State University
Github Skills (22)
c-language10
edn10
python10
machine-learning10
inference10
deep-learning10
tensorflow10
deep-neural-networks10
ai10
neural-network10
cprogramming-language10
optimization10
performance-monitor9
performance-analytics9
performance-measurement9
Programming languages (2)
Github contributions (5)
intel/ai-reference-models
May 2021 - Nov 2022
Intel® AI Reference Models: contains Intel optimizations for running deep learning workloads on Intel® Xeon® Scalable processors and Intel® Data Center GPUs
Role in this project:
ML Engineer
ML EngineerContributions:9 commits in 1 year 7 months
Contributions summary:Kanvi primarily contributed to the image segmentation models within the repository, specifically 3D-Unet for the BraTS dataset. Their work involved optimizing the TensorFlow graphs, improving performance and accuracy, and adding support for int8 and bfloat16 precision. The contributions spanned modifications to both the accuracy and performance scripts, and also included setting up the necessary environment variables for model execution, demonstrating a focus on model deployment and optimization.
optimizationsprocessorstensorflowzoomodel-zoo
tensorflow/tensorflow
Aug 2021 - Sep 2022
An Open Source Machine Learning Framework for Everyone
Role in this project:ML Engineer
ML EngineerContributions:1 review, 16 commits, 23 PRs in 1 year 1 month
Contributions summary:Kanvi implemented a new pattern for fusing operations within the TensorFlow framework, specifically targeting the `_MklFusedBatchMatMulV2` operation. This involved modifying the remapper to identify and combine batch matrix multiplications with preceding multiplication operations. Additionally, the user added tests and addressed review comments, ensuring the functionality and correctness of the new fusion pattern. These changes contribute to optimized performance within the TensorFlow 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.