Sanjay Subramanian

Sanjay Subramanian works on machine-learning methods at the intersection of natural language processing and computer vision, including multi-hop visual reasoning and spatial reasoning from language supervision. His research includes work on generating listener facial responses from speech and on evaluating whether neural networks’ internal reasoning steps are faithful to their predictions.

Career

Cosigned byLoading...

CosignedSign in to view

Based in: Berkeley

Google

May 2018 — Aug 2018

Software Engineering Intern

Akuna Capital

Jun 2016 — Aug 2016

Software Development Intern

University of California - Berkeley

PhD

University of Pennsylvania

Bachelor's studies in software

Products & media

www2.eecs.berkeley.edu

Spatial Reasoning from Language Supervision

A Berkeley technical report on using language-supervised models for spatial reasoning, visual question answering, pose estimation, and medical-image segmentation.

Writing