Hi, I’m Sabera Talukder!
Foundation models should understand the world the way humans do—that’s what I work on.
I have an interdisciplinary background in machine learning, neuroscience, and electrical engineering, which shapes my research on foundation models for messy spatiotemporal data, including:
- the first tokenized time series foundation models,
- neuroscience foundation models (ICLR Oral), and
- multimodal representation learning for vision and audio.
I received my PhD from Caltech on the Rudiments of Next Generation Foundation Models, where I was advised by Yisong Yue. I'm honored to have been an NSF Graduate Research Fellow, a Chen Fellow, and a PIMCO Fellow. I also spent several years helping teach Management of Technology alongside Richard Murray, Stu Feldman, and Eric Schmidt.
Previously, I double-majored with Honors in both EE & Biochemistry at Stanford, where I worked with Juan Rivas and Tom Clandinin.
Before college, I designed water purifiers for homeless children for which I was honored as 1 of 5 Global Google Science Fair Finalists. I'm fond of synchronized swimming: I won US National Gold, US National Silver, and California State Gold medals back in the day.