Nikhila Ravi

Timeline

1. 2016–2017

   Cambridge and a Harvard year

   After BEng and MEng degrees with First Class Honours at the University of Cambridge, she spends 2016–2017 at Harvard on a John F. Kennedy Scholarship, taking computer science classes at Harvard and MIT and working on AI projects in computer vision. She then applies to Facebook and gets a job there.

2. July 2020

   The PyTorch3D library

   In July 2020, she is first author of the paper Accelerating 3D Deep Learning with PyTorch3D, which introduces an open-source library of modular, efficient, differentiable operators for 3D deep learning. It includes a fast, modular differentiable renderer for meshes and point clouds.

3. April 2023

   Segment Anything and SA-1B

   In April 2023, she is a co-author of Segment Anything, which introduces a promptable segmentation model and a dataset of over 1 billion masks on 11 million images. She co-leads the open source Segment Anything project at Meta, covering the Segment Anything Model (SAM) and the SA-1B dataset.

4. 2024

   Young Engineer of the Year

   In 2024, the Royal Academy of Engineering names her one of five winners of its Engineers Trust Young Engineer of the Year competition. She is then a Research Engineering Manager at Meta, leading an AI research engineering team on computer vision in the Fundamental AI Research group (FAIR).

5. August 2024

   SAM 2 for video segmentation

   In August 2024, she is lead author of SAM 2: Segment Anything in Images and Videos, a transformer model with streaming memory for real-time video processing. SAM 2 reaches better video accuracy with 3x fewer interactions than prior approaches and is 6x faster than SAM on images.

6. 2025

   ICLR 2025 honorable mention

   SAM 2 receives one of three Honorable Mentions among the ICLR 2025 Outstanding Paper awards, which also name three outstanding papers. Nikhila Ravi is the first author listed.

7. November 2025

   SAM 3 and concept prompts

   In November 2025, she is among the authors of SAM 3: Segment Anything with Concepts, a unified model that detects, segments and tracks objects in images and videos from concept prompts such as short noun phrases. The team open sources SAM 3 along with the SA-Co benchmark. In December 2025, she appears on the Latent Space podcast as SAM lead at Meta.

8. As of October 2026

   Director at Meta Superintelligence Labs

   She is a Director at Meta Superintelligence Labs, where she co-leads the Segment Anything team and has experience leading research and engineering organizations of 40+ people. SAM 3 is accepted to ICLR 2026, and the SAM 3.1 upgrade doubles video processing speed from 16 to 32 frames per second on a single H100 GPU with no loss in accuracy.