Projects
A collection of my work in computer vision, ML inference, and GPU programming for medical imaging and surgical robotics.
Featured Projects
Ultra-low latency 4K stereo surgical endoscopy pipeline with real-time depth estimation on the edge. I spent a lot of time prototyping with the Holoscan SDK on NVIDIA IGX Orin and Jetson Orin (including Thor), focusing on end-to-end GPU throughput: zero-copy capture/ingest, NVDEC/NVENC, CUDA pre/post-processing, TensorRT inference, and tightly-synchronized stereo processing for consistent depth. Iterated heavily on profiling, memory movement, and scheduling to keep latency predictable for safety-critical medical imaging. Image source: https://nvidia-holoscan.github.io/holohub/workflows/ai_surgical_video/
Diffusion-based video world foundation models for surgical simulation. Features multimodal controls for camera trajectory generation and data augmentation for training surgical AI systems.
Comprehensive multi-endoscope dataset designed for surgical 3D perception research. Enables sub-pixel reprojection accuracy for advanced depth estimation and scene understanding.
More Projects
Innovative auto-focus system using eye-tracking technology with Holoscan SDK. Processes 120fps video streams for responsive, natural focusing behavior in surgical endoscopes.
Offline, voice-enabled Vision Language Model RAG system designed for real-time usage on edge devices. Prioritizes privacy while enabling natural language interactions.
High-resolution endoscopic video dataset for robotic assisted procedures. Designed to advance super-resolution research for surgical imaging (MICCAI 2025 Oral).
4K stereo surgical endoscope prototype development with real-time video processing on NVIDIA Jetson Orin. Integrated with Da Vinci 5 & Xi systems for clinical evaluation.
Cloud-based telehealth data pipeline for patient treatment tracking and anomaly detection. Built with Apache Airflow, dbt, Databricks, and Snowflake.