• Prototype short-form analog and digital games modeling recommendation algorithms, echo chambers, and algorithmic bias.

• Run weekly playtests and design iterations within a 5-person interdisciplinary research team on algorithm literacy systems.

• Contribute toward research demos and submissions for IDC and Play Make Learn conferences.

● Developed a multi-level reinforcement learning environment in IsaacGym for quadcopter navigation and obstacle avoidance tasks.

● Integrated VLM-based control, distilled programs, and weak supervision pipelines for hierarchical policy learning.

● Trained and evaluated PPO agents on forward-avoid and navigation tasks with randomized obstacles and depth-camera input.

● Automated multi-trial experiments and performance visualization (rewards, crashes, successes) to analyze agent behavior.

● Developed an ML-driven ultrasound guidance workflow that learns clinician-defined waypoints and camera trajectories, replaying them with real-time ghost overlays to enable non-experts to reproduce expert scan paths with high fidelity.

● Built a single-camera perception pipeline using computer vision + OCR models (OpenCV, MediaPipe, custom digit-recognition) to perform real-time landmark detection, point verification, and adaptive scan navigation.

● Engineered an ultrasound reconstruction pipeline that algorithmically stitches sequential 2D frames into a coherent 3D model, enabling downstream ML analysis and improved diagnostic visualization for clinicians.