A Site for All Things Milan Shah
Welcome to my page!
Some things about me…
My name is Milan Shah and I am currently a Ph.D. student at North Carolina State University under the direction of Dr. Michela Becchi. In addition to my work at NCSU, I have been working with Argonne National Laboratory under the supervision of Dr. Franck Cappello, Dr. Sheng Di, and Dr. Xiaodong Yu. My research interests lie at the intersection of hardware acceleration, AI/ML, compression, and high performance computing. Everything from writing CUDA code to running large-scale simulations to exploring novel AI techniques excite me, so if any of these topics excite you, please reach out!
Specific topics I deal with…
- Compression algorithms on emerging AI chips, including those from:
- Using lossy compression to reduce the memory footprint of graph neural networks
- GPU-based lossy compression algorithms (branching off the SZ compressor family)
- Integrating compression with tensor network quantum circuit simulators (like QTensor)
- Accelerating random forest classification
- AI/ML algorithms in general!
Questions I’d like answers to…
- How can we make AI/ML more resource-efficient?
- In what ways can we leverage novel architectures to meet exponentially growing computational needs?
- What place does compression have as a tool to reduce power/storage/time requirements?
Some additional thoughts…
I encourage you to explore the rest of the site (such as the Publications page) to see more about some means of addressing these questions. You’ll find papers we have published, code repositories ready to be forked, and even a few blog posts to make my research space more accessible. Thanks for tuning in!