Cyngn

Team Lead Localization • May 2017 - Present

Implemented a pipeline for Simultaneous Localization and Mapping (SLAM) using visual odometry, inertial measurements and map information. Worked on a complete computer vision system for mono and stereo camera, which included the determination of both intrinsic and extrinsic parameters (camera calibration), feature detection and tracking, outliner detection and non-linear estimation.

Developed a Bayesian Filter to combine different sensor information to obtain the optimal pose estimate of a vehicle.

Udacity

Self-Driving Car Nanodegree Program Mentor • Dec. 2016 - Oct. 2017

Mentored students in the Udacity's Classroom. Provide guidance in the area of computer vision and deep learning, with emphasize on Convolution Neural Networks, Behavior Cloning using TensorFlow and Keras.

Future Mobility USA Corp.

Autonomous Driving Localization Systems Lead • Aug. 2016 - May 2017

Formulated and implement localization and mapping algorithms to enable Level 4 autonomous driving. Responsible for building highly efficient, large-scale, distributed data processing pipeline for mapping and localization. Determine platform software requirements and architecture, and decide on the functional components and commendable features for an autonomous driving system.

Identified new technologies that were brought into the team to provide new innovations.

Mercedes Benz Research & Development North America

Software Engineer, Autonomous Driving • Aug. 2015 - Aug. 2016

Responsible for advanced research topics in the area of localization for self-driving cars. Developed efficient and robust algorithms for localization, combining novel tracking techniques with stochastic filtering and graph optimization methods. Designed algorithms such as visual inertial odometry, dead reckoning, map matching and data association for mapping and localization. Evaluated different loosely and tightly coupled GPS/IMU systems and integrated an Interacting Multiple Model (IMM) for a more accurate state estimation.

Represented the company at various conferences and meetings.

Mentored junior engineers on best practices and the current state of the art in the field.

Buffalo University

Postdoctoral Fellow • Aug. 2016 - May 2017

Addressed the problem of fast emulator construction by developing novel strategies for Big Data from computationally expensive simulations. Used a combination of efficient sparse representations of simulation “data” with graph theory, low-rank approximation and multilevel-multiscale methodologies. Improved the Gaussian Process Regression limitation regarding memory requirements and computational demands.

Used Bayesian Model Averaging (BMA) to predict the probability density function (PDF) of the quantity of interest to be predicted and forecasted.

Served as a Co-Principal Investigator in a series of proposal submission including Partnerships for Innovation: Accelerating Innovation Research - Technology Translation (PFI: AIR-TT).