Research

Research Experience

  1. Sep 2025 – Present

    MIT Media Lab

    Multisensory Intelligence Group

    Graduate Researcher

    • Working on a new research direction in Vision-Language-Action (VLA) models and robotics.

  2. Jun 2025 – Sep 2025

    MIT Media Lab

    Multisensory Intelligence Group

    Graduate Visiting Researcher

    • Co-developed PAGE-4D, a feedforward 4D perception framework extending VGGT with a dynamics-aware aggregator for static–dynamic disentanglement.
    • Introduced mask-guided attention to suppress motion for pose tokens while exploiting dynamics for geometry tokens.
    • Applied selective fine-tuning on the middle 10 VGGT layers (~30% parameters), matching full fine-tuning performance with no runtime or memory overhead.
    • Achieved state-of-the-art results on Sintel, DyCheck, and TUM benchmarks, improving depth, pose accuracy, and rendering quality (PSNR/SSIM, LPIPS).

  3. Jul 2024 – Oct 2024

    Imperial College

    Department of Mechanical Engineering

    Undergraduate Researcher, Full-Stack Developer

    • Designed and developed a modern, data-driven web platform called Smart-Forming that enables engineers to discover, evaluate, and share manufacturing knowledge modules.
    • Focused on intuitive UX, modular architecture, and seamless integration of metadata analytics (heatmap, word cloud supported by Python and MATLAB) to support industrial R&D.

  4. Jan 2022 – Dec 2022

    UESTC

    School of Computer Science and Engineering

    High School Researcher

    • Research on Open World Object Detection for classifying known and unknown objects.
    • Improved a Detectron2-based model ORE for incremental object detection using contrastive clustering and auto-labeling RPN.