Vaishaal Shankar

About

I am a research scientist with the Machine Learning Research group at Apple. I am broadly interested in the foundations of machine learning. My work has recently been focused on the interplay between dataset design, multi-modality, and distributional robustness. I have been fortunate to be a part of the ImageNetV2 and OpenCLIP projects.

Previously, I spent 9 wonderful years at UC Berkeley and had the pleasure of working with Ben Recht, Ludwig Schmidt, Eric Jonas, Shivaram Venkataraman, and many others.

Contact me at vs at vaishaal dot com.

Selected Publications

• Do ImageNet classifiers generalize to ImageNet? - B Recht, R Roelofs, L Schmidt, V Shankar, ICML 2019
• Data Determines Distributional Robustness in Contrastive Language Image Pre-training (CLIP) - A Fang, G Ilharco, M Wortsman, Y Wan, V Shankar, A Dave, L Schmidt, ICML 2021
• Measuring Robustness to Natural Distribution Shifts in Image Classification - R Taori, A Dave, V Shankar, N Carlini, B Recht, L Schmidt, NeurIPS 2020
• Self Supervision Does Not Help Natural Language Supervision at Scale - F Weers, V Shankar, A Katharopoulos, Y Yang, T Gunter, CVPR 2023