2019-2020: Primary engineer for highway driving efforts which increased the internal driving metrics by approximately 3 times in 7 months. This was driven by improvements in lane changes, merges, cut-ins, and collaborations with the mapping, perception, and prediction teams.2020-current:Currently leading the junctions group which is a large cross functional team consisting of engineers from perception, planner, prediction, QA, and simulation. My role involves having expert and holistic knowledge of all junction issues, mentoring several engineers to help develop algorithms that are both theoretically solid and work well in practice, helping define scope of company milestones where I provide insights needed to make key tradeoffs between engineering complexity and timelines, setting a technical vision for the team, creating processes and structures to make the team more efficient, and leading several complex engineering projects of my own.I'm also part of a pilot program offered to the top 5% of performers in the company which provides leadership coaching and accelerated development opportunities.

Developed a junction handling framework that enables self-driving cars to drive safely and legally through arbitrary protected and unprotected intersections. The framework was tested and deployed in both US and China, and is currently being used on public roads for investor demos. Developed algorithms to determine occluded areas of the map based on laser point cloud data, and designed behaviors to overcome the missing information and make progress safely.Traveled to China and supported the local team in adapting, expanding, and testing the AI stack.

I was involved with two projects during my time here. Online learning and planning for scientific exploration:Scientists involved in rover missions are often interested in variables that require specialized sensors to measure. However, deploying these sensors is usually either energetically expensive or requires the robot to remain stationary. A new method to automatically learn correlations between these expensive to use sensors and the other on-board sensors such as cameras on-line was developed which improves the rate at which scientific information is gathered. Applications to a NASA rover mission which aims to mine surface volatiles on the moon was explored. The work was presented and published at the Field and Service Robotics conference and in Autonomous Robots journal. Informative trajectory planning for autonomous spacecraft entry, descent and landing: Large communication delays make landing on planetary bodies is a challenging robotics problem. Due to sensing limitations, essential geological features which determine safety and science utility of a landing site can’t be perceived accurately from orbit. A sampling based sensing and motion planning technique was developed which can automatically plan informative lander trajectories in real time. This ensures the lander can gain enough data to determine with high confidence that the landing site is suitable for the mission goals. The research was presented and published at the AIAA Space Forum, and iSAIRAS, which are two leading venues for space robotics research.

Taught undergraduates engineering mechanics and supervised labs in which they learned how to create a simple Arduino program to interact with hardware.

Designed and carried out experiments to benchmark the performance of several prototypes of a portable respiratory humidification machine to be used in a hospital setting. Also built a pneumatic based testing rig to help characterize the long term deformation properties of polymer mixtures when subjected to various levels of loading.

Worked with a PhD student in the design of a five degree of freedom upper limb exoskeleton for stroke rehabilitation. Main responsibility was to develop low cost 2-axis force sensors along with a hand interface such that key variables indicating the patient’s performance could be measured and used to adapt the exoskeleton’s stiffness.