• Led end-to-end camera sensor bring up, system integration and V&V

• Developed in house image processing library on CUDA from scratch eliminating need for third party libraries.

• Optimized image capture to image consumption pipeline for target GPU/CPU architecture enabling 6x performance jump in processing throughput.

• Built and added numerous bespoke tools for data extraction, data visualization and replay in C++, used by different teams and automated integration tests.

• Implemented ML inference engine deployment pipeline.

• Feature architect of over-the-air firmware update on STM microcontroller ECU.

• Worked on deep learning based interaction primitives to make exoskeletons adaptive

• Toolchain : Python 3.0, Tensorflow 2.0, Tensorflow Probability, Keras, NumPy

• Recieved MORE Research Grant at ASU (Project abstract: https://furi.engineering.asu.edu/furiproject/learning-deep-neural-interaction-policy-for-exoskeleton-control/ )

• Recieved Gerald Ferin Memorial Fellowship at ASU

Advisor : Dr. Heni Ben Amor

• Designed and Developed quadrupedal walking robot Stoch 2

• Developed high fidelity sim-to-real transfer framework for Stoch 2

• Stoch 2 demonstration at ICRA 2019 Learning Legged Locomotion Workshop