At Agility Robotics, I develop C++ and Python robotics software and integrate 3D perception pipelines into real-time ROS 2 workflows. I optimized model inference with ONNX and TensorRT, reducing latency by approximately 30% while maintaining acceptable perception accuracy.
I also improved perception throughput from approximately 15 FPS to 30 FPS and helped achieve an approximately 10% improvement in mAP@50 through model tuning and inference optimization. I use camera, RGB-D, LiDAR, and IMU data to support perception and sensor processing.
Previously, I supported robotics software and perception development as a Robotics Engineer Intern at Figure AI, and developed production machine learning solutions at Kore.ai. My work there included NLP and transformer-based solutions, model evaluation, and FastAPI inference services.
