Spandan Tiwari
Redmond, Washington, United States
Summary
🤩
RockstarSpandan Tiwari is Director of Engineering for AI Frameworks and Applications at AMD AI, leading strategy and execution for framework-level capabilities that accelerate GenAI, LLMs, computer vision, ASR and NLP workloads across Ryzen AI NPUs and Instinct MI300x GPUs. He specializes in connecting major ecosystems—PyTorch, ONNX/OnnxRuntime and Hugging Face—to cutting-edge accelerators via compilers, runtimes, graph optimizers and advanced quantization/runtime features such as distributed execution. Previously he led the PyTorch@Meta compiler/runtime effort for accelerator inference/training and was a founding contributor to the ONNX standard at Microsoft, with hands-on contributions to ops like Flatten, Conv and OneHot in the onnx repo. PhD-trained with roughly two decades of experience spanning computational photography, image processing and ML systems, he also led research that earned a U.S. Army SBIR Achievement Award. Known for product-oriented research and team building, he combines deep algorithmic expertise with mentoring and roadmap-driven delivery.
9 years of coding experience
Github Skills (10)
mle10
machine-learning10
deep-learning10
onnx10
ml10
python9
neural-networks9
pytorch8
tensorflow28
tensorflow8
Programming languages (5)
Github contributions (5)
onnx/onnx
Jan 2018 - Jan 2019
Open standard for machine learning interoperability
Role in this project:
ML Engineer
ML EngineerContributions:2 reviews, 11 commits, 23 PRs in 11 months
Contributions summary:Spandan contributed significantly to the ONNX repository by implementing and testing new operations (ops) for machine learning models. They developed tests for the Flatten, Conv, ConstantLike, EyeLike, and OneHot ops. The user also contributed to the definition of these ops and updated relevant files. Their work involved adapting the ONNX framework to support a wider range of machine learning models.
pytorchmxnetdeep-learninginteroperabilitymachine-learning
spandantiwari/pytorch
Sep 2018 - Jan 2021
Tensors and Dynamic neural networks in Python with strong GPU acceleration
Contributions:121 pushes, 46 branches in 2 years 4 months
pythongpu-accelerationdeep-learninggpuacceleration
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.