2025: Generalized Neighborhood Attention: Multi-dimensional Sparse Attention at the Speed of Light (Preprint); 2024: Faster Neighborhood Attention: Reducing the O(n^2) Cost of Self Attention at the Threadblock Level, NeurIPS 2024; 2023: Neighborhood Attention Transformer, CVPR 2023; 2022: Dilated Neighborhood Attention Transformer (Preprint).
Research Experience
Research Scientist at NVIDIA Research, Deep Imagination, since 10/2025; Graduate Researcher at SHI Labs, Georgia Tech, 01/2024 - 10/2025; Research Intern at NVIDIA Research, Deep Imagination, 12/2024 - 07/2025; Software Performance Engineering Intern at NVIDIA, CUTLASS, 05/2024 - 08/2024; Software Engineering Intern at HippoML, 06/2023 - 12/2023; Graduate Researcher at SHI Labs, University of Oregon, 03/2021 - 12/2023; Research Intern at Picsart AI Research, 06/2022 - 09/2022; Undergraduate Researcher at University of Kerman, Mahani Mathematical Research Center, 2019 to 2021.
Education
PhD in Computer Science, 2024 - 2026 (expected), Georgia Tech, advised by Prof. Humphrey Shi; M.S. in Computer Science, 2021-2023, University of Oregon, advised by Prof. Humphrey Shi; B.S. in Computer Science, 2016-2020, University of Kerman, research advisor: Prof. Abbas Salemi.
Background
A Research Scientist at NVIDIA's Deep Imagination Research Group. My primary research area is High-Performance AI Architecture, which involves making fast and efficient AI, from neural network architecture down to computer architecture. I've been working on Attention and Sparse Attention since about 2022.
Miscellany
Links to GitHub, Google Scholar, Twitter, LinkedIn, etc.; Teaching experience includes TA for CS 4476: Intro to Computer Vision at Georgia Tech in Spring 2025, Instructor for CS 322: Intro to Software Engineering at University of Oregon in Fall 2023, and TA for CS 322: Intro to Software Engineering at University of Oregon in Spring 2021, Winter 2022, and Winter 2023.