Mit Kotak
Scholar

Mit Kotak

Google Scholar ID: DFr2AOcAAAAJ
MIT
domain specific languagesgeometric deep learning
Citations & Impact
All-time
Citations
33
 
H-index
3
 
i10-index
1
 
Publications
8
 
Co-authors
17
list available
Resume (English only)
Academic Achievements
  • “The Price of Freedom: Exploring Tradeoffs in Equivariant Tensor Product Operations”, ICML 2025 (co-author)
  • “High-performance training and inference for deep equivariant interatomic potentials” (co-author)
  • MS Thesis (2025): “Tensor Products in E(3) Equivariant Networks”
  • “Simplifying Equivariant GPU Kernels through Tile-based Programming”
  • “Efficiently executing data flow graphs on GPUs”, CEESD AST Review 2022 (co-author)
  • “Efficiently executing Numpy on GPUs via the CUDAGraph API”, UIUC URS 2022 (co-author)
  • Grant progress report: “3D Visualizations of Black Hole Binaries and Disks in Full General Relativity”
  • “CyberGIS-Compute: Middleware for Democratizing Scalable Geocomputation” (SSRN)
  • “Streamlined HPC Environments with CVMFS and CyberGIS-Compute”, iGuide 2023
Research Experience
  • Research at MIT with advisors Tess Smidt and Saman Amarasinghe
  • 3D visualization of astrophysics simulations with Prof. Stuart Shapiro at Illinois Relativity Group
  • HPC middleware for geospatial computing with Prof. Anand Padmanabhan at CyberGIS Center
  • High-performance computing for neutron stars with Prof. Antonios Tsokaros at National Center for Supercomputing Applications
  • GPU task graph parallelism via CUDAGraphs with Prof. Andreas Klöckner at CEESD