I’m a Research Scientist at FAIR, part of Meta Superintelligence Labs, working on agents and computer vision. My group works on two main areas: computer use agents, most recently the CUA capabilities in Muse Spark 1.1 and Muse Glimmer, and a new initiative on agentic capability for physical engineering, teaching models to design real hardware such as robotics, vehicles, and semiconductors. Before FAIR I was at Amazon AWS AI where I was part of the team that launched Rekognition Video. and before that Amazon GO. I got my PhD under the supervision of Svetlana Lazebnik at UNC Chapel Hill with a focus on nonparametric methods for semantic segmentation.

News

• [2026] Muse Glimmer is out, an open agentic model that runs on device, with computer use capability from our team blog
• [2026] I’ll be on the Senior Program Committee for AAAI 2027
• [2026] I’m a Lead AC for CVPR 2026 and an AC for both NeurIPS 2026 and ECCV 2026
• [2026] Hierarchical Procedural Meta-Reasoning for Generalizable Multimodal Agents was accepted to the ICLR workshop on Agentic AI in the Wild pdf
• [2026] Muse Spark 1.1 is out with strong Computer Use Agent capabilities, about two years of work from the team blog
• [2026] Our paper on injecting 3D spatial priors into vision language models for geometric reasoning was accepted to CVPR pdf
• [2026] Enhancing Visual Planning with Auxiliary Tasks and Multi-Token Prediction was accepted to WACV pdf
• [2026] Computer Use at the Edge of the Statistical Precipice is now on arXiv pdf
• [2025] I’m an AC for CVPR 2025 and NeurIPS 2025
• [2025] Our wearable agent goal inference benchmark was a Spotlight (~3%) at the NeurIPS 2025 Datasets and Benchmarks track pdf code
• [2025] DigiData: Training and Evaluating General-Purpose Mobile Control Agents pdf site
• [2025] ADEPTS: A Capability Framework for Human-Centered Agent Design pdf
• [2025] Embodied AI Agents: Modeling the World pdf
• [2025] My group is hiring research scientists and interns to work on digital and wearable agents, reach out if you are interested
• [2024] I’m an AC for CVPR 2024 and NeurIPS 2024
• [2024] Threshold-Consistent Margin Loss for Open-World Deep Metric Learning was accepted to ICLR pdf
• [2024] Learning for Transductive Threshold Calibration in Open-World Recognition was accepted to CVPR pdf
• [2024] Benchmarking Zero-Shot Recognition with Vision-Language Models was accepted to the CVPR workshop on Multimodal Foundation Models pdf
• [2023] I’ve been asked to be an AC for NeurIPS 2023.
• [2023] SkeleTR: Towards Skeleton-based Action Recognition in the Wild was accepted to ICCV pdf
• [2023] ScaleDet: A Scalable Multi-Dataset Object Detector was accepted to CVPR pdf
• [2022] I’ll be one of the ACs for the upcoming CVPR 2023.
• [2022] An In-depth Study of Stochastic Backpropagation was accepted to NeurIPS pdf
• [2022] Our new tracking dataset is now available here!
• [2022] We had 3 papers accepted to ECCV Links Coming soon
• [2022] We had 4 papers accepted to CVPR
• [2021] Our NUTA paper was accepted to WACV
• [2021] We have an upcoming challenge on tracking Airborne Objects at ICCV 2021 site
• [2021] 3 papers accepted to ICCV!
  • VidTr: Video Transformer Without Convolutions pdf
  • Single View Physical Distance Estimation using Human Pose pdf
  • Selective Feature Compression for Efficient Activity Recognition Inference pdf
• [2021] Presented a tutorial on Leveraging Motion in Video at CVPR talk site
• [2021] SiamMOT paper was accepted to CVPR pdf
• [2020] Check out the video models now available in the Gluon pytorch tool kit my team helped develop link