Alok Tripathy

Member of Technical Staff

Essential AI

Biography

I’m a Member of Technical Staff at Essential AI, where we are pushing the capabilities of fully open-source LLMs.

I recently graduated with a CS Ph.D. from UC Berkeley and Lawrence Berkeley National Lab. My dissertation focused on scaling large GNN models and applying them to sciences, particularly high-energy physics. I was very thankful to be advised by the wonderful Aydın Buluç and Kathy Yelick during this time. In the past, I worked on accelerating graph analytics on GPUs with Oded Green and David Bader at Georgia Tech. My work has been supported by the NSF Fellowship.

Interests

Parallel/Distributed Computing
Machine Learning
Scientific Computing
Graph Analytics

Education

Ph.D. in Computer Science, 2025

UC Berkeley
B.S. in Computer Science, 2019

Georgia Tech

News

“Sparsity-Aware Communication for Distributed Graph Neural Network Training” accepted to ICPP 2024

Jun 2024

Led by my mentee, Ujjaini Mukhopadhyay

“Distributed Matrix-Based Sampling for Graph Neural Network Training” accepted to MLSys 2024

Mar 2024

Passed Ph.D. Qualifying Exam

Feb 2024

“Communication-Avoiding Algorithms for Full-Batch and Mini-Batch GNN Training” accepted to NVIDIA GTC 24

Feb 2024

TA’d for JamCoders 2023

Jun 2023

This is a phenomenal experience. For anybody interested in teaching CS or algorithms, I strongly recommend applying.

Publications

Ujjaini Mukhopadhyay, Alok Tripathy, Oguz Selvitopi, Katherine Yelick, Aydın Buluç (2024). Sparsity-Aware Communication for Distributed Graph Neural Network Training. Proceedings of the 53rd International Conference on Parallel Processing.

PDF

Alok Tripathy, Katherine Yelick, Aydın Buluç (2023). Distributed Matrix-Based Sampling for Graph Neural Network Training. Proceedings of Machine Learning and Systems, arXiv:2311.02909.

PDF

Oguz Selvitopi, Benjamin Brock, Israt Nisa, Alok Tripathy, Katherine Yelick, Aydın Buluç (2021). Distributed-Memory Parallel Algorithms for Sparse Times Tall-Skinny-Dense Matrix Multiplication. ACM Proceedings of the ACM International Conference on Supercomputing (ICS), 2021.

PDF

Alok Tripathy, Oded Green (2021). Scalable Hash Table for NUMA Systems. arXiv:2104.00792.

PDF

Alok Tripathy, Oded Green (2020). Accurately and Efficiently Estimating Dynamic Point-to-Point Shortest Path. IEEE BigGraphs Workshop at International Conference on Big Data (BigData) 2020.

Alok Tripathy, Katherine Yelick, Aydın Buluç (2020). Reducing Communication in Graph Neural Network Training. ACM/IEEE International Conference for High Performance Computing, Networking, Storage and Analysis, arXiv:2005.03300.

PDF

James Fox, Alok Tripathy, Oded Green (2019). Improving Scheduling for Irregular Applications with Logarithmic Radix Binning. IEEE High Performance Extreme Computing (HPEC) 2019.

PDF

Alok Tripathy, Fred Hohman, Duen Horng Chau, Oded Green (2018). Scalable K-Core Decomposition for Static Graphs Using a Dynamic Graph Data Structure. IEEE International Conference on Big Data (BigData) 2018.

PDF

Oded Green, James Fox, Alex Watkins, Alok Tripathy, Kasimir Gabert, Euna Kim, Xiaojing An, Kumar Aatish, David A. Bader (2018). Logarithmic Radix Binning and Vectorized Triangle Counting. IEEE High Performance Extreme Computing (HPEC) 2018.

PDF

Alok Tripathy, Oded Green (2018). Scaling Betweenness Centrality in Dynamic Graphs. IEEE High Performance Extreme Computing (HPEC) 2018.

PDF