Reza Zadeh

Also known as Reza Bosagh Zadeh, رضا زاده

Reza Zadeh is a computer scientist who helped build the machine learning tools in Apache Spark and now runs a computer vision company. After studying at Waterloo and Carnegie Mellon, he finished a Stanford PhD in 2013 and built the machine learning behind Twitter's who-to-follow system. Stanford gave him its 2014 Gene Golub Doctoral Dissertation Award, and his Spark paper on matrix computations won a KDD Best Paper Award (runner-up) in 2016. He founded Matroid in 2016 and wrote TensorFlow for Deep Learning with Bharath Ramsundar, published by O'Reilly in 2018. Today he is founder and CEO of Matroid and an adjunct professor at Stanford.

Now at: Matroid and Stanford University
Works: 50
Citations: 3,738
h-index: 18
Badges: Founder
Bridge
Boomerang
Links: Website and Google Scholar

Publication numbers from OpenAlex, as of 22 October 2026.

Timeline

2004–2013

From Waterloo to a Stanford PhD

He studies for a Bachelor of Mathematics in honours computer science at the University of Waterloo from 2004 to 2008, and works as a Google researcher on statistical machine translation from 2006 to 2007. He then completes a Master of Computer Science at Carnegie Mellon University and, in 2013, a PhD in computational mathematics at Stanford under Gunnar Carlsson, with the thesis Large scale graph completion.
2011–2013

Who to Follow at Twitter

As a senior data scientist at Twitter from 2011 to 2013, he builds the machine learning algorithms behind the who-to-follow system, the first product to use machine learning at Twitter. His WWW 2013 paper WTF: The Who to Follow Service at Twitter describes the service.
2014

A Golub dissertation award

Stanford's Institute for Computational & Mathematical Engineering gives him the 2014 Gene Golub Doctoral Dissertation Award for his thesis Large scale graph completion.
2015

MLlib in Apache Spark

He is among the 16 authors of the 2015 paper MLlib: Machine Learning in Apache Spark, which presents Spark's open-source distributed machine learning library. He is co-creator of MLlib and of the Linear Algebra Package in Apache Spark.
2016

A KDD runner-up paper award

A KDD 2016 Best Paper Award (runner-up) goes to Matroid for Matrix Computations and Optimization in Apache Spark, a collaboration with researchers at Databricks, HP Labs, Twitter, UC Berkeley, MIT and Stanford. He is the first author of the paper, which covers distributed matrices and optimization routines in Spark, including a Spark port of the TFOCS optimization package.
2016–2017

Founding Matroid

He founds Matroid in 2016 to commercialize computer vision research. In September 2017, the company announces a $10 million Series A co-led by NEA and Intel Capital, which brings its total funding to $13.5 million.
2018

TensorFlow for Deep Learning

He and Bharath Ramsundar write TensorFlow for Deep Learning, which O'Reilly publishes in 2018.
As of October 2026

Matroid CEO and Stanford adjunct

In June 2024, he represents Matroid on the sidelines of the G7 Summit at the Vatican and meets Pope Francis to discuss artificial intelligence. He is founder and CEO of Matroid, a computer vision company whose users build and deploy detectors without writing code, and an adjunct professor at Stanford, where he teaches Distributed Algorithms and Optimization (CME 323) and Discrete Mathematics and Algorithms (CME 305).