Mahesh Sathiamoorthy

Staff Software Engineer at Google Brain

Brief Bio

I am Maheswaran Sathiamoorthy, but go by Mahesh Sathiamoorthy.

I am a Software Engineer at Google since 2014 and work on ML/Recommender Systems/TPUs (since 2017). From 2008 to 2013, I was a PhD student at Department of Electrical Engineering, University of Southern California. I worked with Prof. Bhaskar Krishnamachari and Prof. Alex Dimakis.

Before that, I did my B.Tech(H) in Electronics and Electrical Communication Engineering at the Indian Institute of Technology Kharagpur, India.

Contact me at mahesh at smahesh dot com.

Research/Current Work

I am a Tech Lead in Google Brain and currently work on large scale real world recommender models, with a goal of scaling them up to harness the benefits of large models. I work on all aspects of recommender models: infrastructure and productionization, research and engagements. My work benefits surfaces such as YouTube.

Selected Publications

  1. Kiran Vodrahalli, Rakesh Shivanna, Maheswaran Sathiamoorthy, Sagar Jain, Ed H. Chi. "Algorithms for Efficiently Learning Low-Rank Neural Networks". 2022 (preprint)
  2. Hussein Hazimeh, Zhe Zhao, Aakanksha Chowdhery, Maheswaran Sathiamoorthy, Yihua Chen, Rahul Mazumder, Lichan Hong, Ed H Chi. "DSelect-k: Differentiable Selection in the Mixture of Experts with Applications to Multi-Task Learning". NeurIPS 2021 [arxiv]
  3. Zhe Zhao, Lichan Hong, Li Wei, Jilin Chen, Aniruddh Nath, Shawn Andrews, Aditee Kumthekar, Maheswaran Sathiamoorthy, Xinyang Yi, and Ed Chi. 2019. "Recommending what video to watch next: a multitask ranking system". In Proc. of the 13th ACM Conference on Recommender Systems (RecSys '19). ACM, New York, NY, USA, 43-51. DOI: [Google]
  4. Stephen Macke, Alex Beutel, Tim Kraska, Maheswaran Sathiamoorthy, Derek Zhiyuan Cheng, Ed H. Chi. Lifting the Curse of Multidimensional Data with Learned Existence Indexes. ML for Systems workshop at NeurIPS, 2018. [pdf]
  5. Maheswaran Sathiamoorthy, Megasthenis Asteris, Dimitris Papailiopoulos, Alex G. Dimakis, Ramkumar Vadali, Scott Chen, Dhruba Borthakur, "XORing Elephants: Novel Erasure Codes for Big Data", VLDB 2013.

Blog posts

An introduction to JAX

What is JAX? JAX is NumPy but more with various functionalities designed to make machine learning research faster. It introduces a functional programming paradigm and has other valuab...

In tech, Apr 04, 2021


I started a new blog called at the beginning of the year.

In blog, Apr 04, 2021

2020: It's a wrap!

Looking back at my blog, I am disappointed that my last post was in 2018. In fact, my 2018 resolution was to write 50k public words, but I barely got to a few thousand.2019 seems to h...

In blog, Dec 29, 2020

Finding Passion

One would think that Jeff Bezos’s passion has always been Amazon and online retail, but turns out his passions areRockets, space travel and propulsion.

In mind over matter, Jul 11, 2018

Dark Knowledge and distillation

Guess what this image contains?

In tech, May 27, 2018

Surely You're Joking, Mr. Feynman!

I recently finished this book (should have read it long back!) and I must say, what a great read about a great and a curious mind!

In mind over matter, Jan 04, 2018

New Year's resolutions

2018 is actually here (and that’s bad because it looks like time is crunching through years faster than I would like). And that also means it is time to set new year’s resolutions.

In mind over matter, Jan 01, 2018

Understanding TensorFlow Graph Execution with a Simple Example

This post is for absolute beginners. I hope to be able to explain complicated concepts in simple terms to benefit a wider audience (see dummies guide to erasure coding post for a diff...

In tech, Jul 10, 2017

Procrastination Filter

Have you ever noticed drowning into a sea of articles that you opened via Hacker News? Did you end up buying those nine little things from Amazon of which you don’t use any? If you t...

In mind over matter, Oct 14, 2015

Popular Weekend Programming Languages

What are some languages used most often during the weekends? Are there some languages that are inherently more ‘hobbyist’ than others?

In tech, May 27, 2013