I’m a Robotics Engineer and Researcher with experience spanning 3D perception, embedded ML acceleration, and full-stack autonomy system design. Currently pursuing my M.S. in Mechanical Engineering at Carnegie Mellon University, I work on developing high-fidelity registration, auto-calibration, and real-time perception toolboxes for robotic systems.
My background bridges research and deployment:
At CMU, I’ve built toolchains for sub-millimeter registration accuracy, real-time object tracking (YOLOv11, SAM2, CoTracker v3), and autonomy speedups of 30×.
At Acceleration Robotics, I accelerated perception pipelines on Qualcomm RB5 and NVIDIA Jetson Orin, achieving 4× speedup with CUDA/OpenCL and optimizing ML models for low-latency, resource-constrained edge platforms.
At EPFL, I worked on bio-inspired UAV modeling and real-time gait energetics estimation, applying ML to complex physical systems.
As an open-source contributor to ROS2 Navigation, I implemented recovery behaviors to improve robustness in real-world navigation.
I’m particularly excited about applying my skills in localization, perception, and embedded ML optimization to help robots operate safely and efficiently in dynamic, semi-structured environments.
Core skills: Robotics Autonomy (ROS/ROS2, Nav2), CUDA/TensorRT, Jetson optimization, PyTorch/ONNX, perception pipelines, high-performance computing, C++/Python.
MOLOCH-dev (Anushree Sabnis) (github.com)
Computer vision, High performance computing, Optimization, Learning for 3D vision
Funded by Graduate Fellowship
August 2024 - PresentProject Presentation: 1 | 4 | 5
October 2023 - August 2024March 2023 - October 2023Funded by EPFL SRP Program
I worked on the development of a model-based optimization framework leveraging aeroelasticity in Aerial-Aquatic Ornithopters.