Experience

Qualcomm logo Qualcomm

May 2026 - Current XR Systems Engineering Intern
  • Run capability evaluation and performance analysis for GPU workloads in VR/AR device contexts.
  • Connect hardware constraints, graphics performance, and XR product requirements into practical system findings.

Tesla logo Tesla

Jan 2026 - May 2026 Computer Vision Intern
  • Applied classical computer vision and ML models to identify manufacturing defects.
  • Created a semi-automated annotation and augmentation pipeline for downstream model training and tuning.
  • Improved ML inference and database interaction latency by roughly 2000x, from 1.4 s to 0.8 ms per image.

Safe Robotics Laboratory logo Safe Robotics Laboratory

Apr 2024 - Present Graduate Researcher
  • Trained large foundation models, including VLMs and VLAs, with 3D perception for embodied navigation.
  • Auto-generated photorealistic indoor 3D scenes for large-scale self-supervised navigation training.
  • Built a humanoid teleoperation client-server system for Meta Quest headsets and MuJoCo simulation.

Georgia Tech Research Institute logo Georgia Tech Research Institute

Jan 2024 - Aug 2024 Embedded Systems Intern
  • Built agentic AI workflows for fighter pilots to request on-demand data visualization through voice interaction.
  • Fabricated, assembled, and documented hardware prototypes for classified DoD projects.

SightWise logo SightWise

Apr 2023 - Dec 2023 Co-Founder
  • Launched an AI visual scene description service that reached more than 1.5k users.
  • Fine-tuned large language models for visual description and deployed GPU-backed services on AWS EC2.
  • Used ONNX and Apache TVM to accelerate mobile computer vision models, reaching a 5x speedup on Android and iOS.

HEART Lab @ Asan Medical Center logo HEART Lab @ Asan Medical Center

May 2022 - Aug 2022 Assistive Technology Research Intern
  • Developed a finger-worn visual-tactile sensory substitution device for human participants with blindness.
  • Implemented object detection, segmentation, tracking, edge detection, visual grounding, and VQA models for real-time edge inference.