Computer Vision & Robotics Simulation Engineer
Physically-based sensor simulation · AI-assisted annotation · Real-time inference on the edge
I build tools and simulations that close the gap between synthetic data and real sensors:
- Sensor simulation — physically-based infrared/thermal cameras in NVIDIA Isaac Sim and Unreal Engine (Planck radiometry, atmosphere, optics, detector noise, AGC).
- Computer vision tooling — GPU-accelerated annotation with SAM 2/3, transformer trackers and open-vocabulary detection.
- Deployment — TensorRT / ONNX pipelines, C++ inference, ROS 2 and low-latency video on embedded hardware.
| Project | What it does | Stack |
|---|---|---|
| TCIsaacSim | Physically-based, per-pixel thermal (IR) camera simulator for Isaac Sim — every pixel carries its own temperature, emissivity and range. Pure-Python physics core is complete and tested; Isaac Sim extension in progress. | Python · Isaac Sim · CUDA · pytest |
| VIAT | Video & image annotation tool with SAM 2.1 / SAM3 / SAM2-TensorRT, OSTrack & E.T.Track propagation, YOLO-World auto-labeling, CLIP zero-shot classification and a built-in mAP evaluation suite. Exports YOLO, COCO, VOC, CreateML. | Python · PyTorch · TensorRT · Qt |
- 🌡️ Taking TCIsaacSim from physics core to a full Isaac Sim sensor (SPG shader, AOVs, ROS 2 output)
- 🌧️ An open-source weather effects extension for Isaac Sim
Languages
Vision & ML
Simulation & Robotics
Open to collaboration on sensor simulation, synthetic data and perception tooling.