About me
My name is Sandeep Polisetty.
I am a fifth year computer science MS/PhD student at the University
of Massachusetts, Amherst.
I am broadly interested in the systems for efficient machine learning with special interest in Graph Neural Networks, Quantization and Compiler optimizations for novel hardware.
I look to attack every point of the stack for machine learning from novel hardware pipelines in verilog to efficient compilers such as XLA and TVM. I am actively seeking research internship
where I can work on related prolems.
I am co-advised by Prof. Hui Guan and Prof. Marco Serafini.
My recent projects are:
- Ongoing project optimizing distributed training of GNN through locality centric execution
- Optimizing sampling on GPUs for Graph Neural Networks. (published: Eurosys 2021)
- polyhedral optimization for dynamic shape operators using TVM. (Presented at TVMConf work done during internship at reservoir labs summer 2021)
Apart from computer science, I love olympic weightlifting (especially snatch) and all outdoor rustic activities such as bushcraft, surfing, scuba diving. Apart from GPUs and communication scheduling, I can talk for hours on these other topics.
Contact me: at sandeep06011991@gmail.com