Hi! I’m Udit Saxena.
I am a machine learning engineer at Sumo Logic with the Advanced Analytics team. I’ve completed my Master’s in Computer Science from UMass, Amherst. My interests lie in the field of Machine Learning - specifically Natural Language Processing and Deep Learning.
At Sumo, I work on building distributed, fast, and approximate streaming clustering algorithms for text streams and time series models. I also work on integrating microservices and running production ML workflows on Kubernetes (AWS EKS) and have worked closely with Airflow, Helm, ArgoCD, Docker and AWS Elastic Container Registry.
My chief research interests lie in sparse approaches to Deep Learning models, where I’ve worked on using Neural Tangent Kernels to estimate a data independent kernel (called the Path Kernel) for the sparse representation of a deep neural network at initialization (paper submitted to ICLR - link to preprint soon).
Previously, I’ve interned with Sumo Logic working on Distributed Tracing in the Bay Area with the metrics team led by David Andrzejewski. Before this I worked as a Product Engineer with the Core team at Sprinklr and have interned with Adobe on User Analytics with the Adobe Captivate team.
In Spring 2018, I worked with the Microsoft Research team at Montreal and Cambridge on Active Learning for transferring knowledge for Reading Comprehension systems with T.J. Hazen. I also interned with the Machine Learning team at Lexalytics where my focus was on Graph Convolutional Networks for Text Classification. I’ve worked on a Pytorch implementation of End to End Memory Networks, which is currently open as a pull request on the official Pytorch examples repository. In Spring 2017, I worked with Prof. Andrew McCallum on serving pre-trained Tensorflow models for the JVM as part of IESL. I was a Google Summer of Code intern, working with MLPACK (along with Ryan Curtin), a C++ based open source machine learning library.
Check out my resume