Pratik Gujjar

I am currently working as a Software Engineer, Machine Learning at Stripe. At Stripe, my work is on building a new and comprehensive ML model training orchestrator. We are also simultaneously exploring a framework that can manage the orchestration of any data-oriented code pipeline packaged as computational graphs.

Previously, I was a Senior Machine Learning Platform Engineer at SceneBox Inc. SceneBox was acquired by Applied Intuition in March 2023. At SceneBox I spent my time on the cutting-edge of how to seamlessly and scalably develop machine learning solutions based on the paradigm of Software 2.0 [1] [2], software that is not written but is learnt from data. Some of my most recent work at SceneBox are streaming data directly from the cloud to the ML training machine Streamable Sets, ML workflow management platform Campaigns and a high-throughput low-latency model-serving platform.

Prior to SceneBox, I was working as a Machine Learning Researcher at Huawei Technologies’ Vancouver Research Centre. At Huawei, I experimented with ML model packaging and serving with Knowledge Distillation, Quantization and Pruning. My favorite bit of work was on a Self-Supervised Semi-Supervised Learning called AuxMix [paper], wherein we make no assumpitions on the data distribution for unlabeled data with respect to labeled data.

I graduated with my MSc degree from, Simon Fraser University. Under the supervision of Professor Richard Vaughan, I defended my thesis work on predicting the near future of pedestrians in video [paper]. Professor Anoop Sarkar and Professor Yasutaka Furukawa were part of the committee that accepted my thesis with no corrections.

I also worked as Senior Technical Associate at the Avaya India Research Centre in Bangalore between 2015 and 2016, where I developed video-conferencing solutions with the Google Glass. I completed my undergraduate degree in Electronics and Communication Engineering from the National Institute of Technology Karnataka, India in 2015 under the supervision of Professor Sumam David.