Headline

AI @ Qualcomm | Robotics @ UPenn

About

I'm a Senior Machine Learning Engineer at the Autonomous Driving team at Qualcomm Research. I focus on Reinforcement Learning and developing Behavior Prediction models for Qualcomm's Autonomous Driving stack. In the past, I've focused on Efficient AI at Qualcomm for the AV stack, working on hardware-software co-design for inference optimization, working on Quantization, Efficient model design, and Neural Architecture Search. I received my master's degree in robotics from the University of Pennsylvania, working in the GRASP Lab. Check out my website https://ramanans1.github.io

Location

San Francisco Bay Area

AI stack

aistudio

cursor

github copilot

openai

Work experience

Senior Machine Learning Engineer at Qualcomm, Nov 2022 - Present · 3 yrs 8 mos, Santa Clara, California, United States, Working in the ADAS ML Systems R&D team

Machine Learning Engineer at Qualcomm, Jun 2020 - Nov 2022 · 2 yrs 6 mos, San Diego, California, United States, Working in the ADAS ML Systems team at Qualcomm Research

Research Assistant at University of Pennsylvania, Dec 2018 - Dec 2020 · 2 yrs 1 mo, Greater Philadelphia, Working with Dr.Kostas Daniilidis in the GRASP Lab on Deep Reinforcement Learning

Teaching Assistant at University of Pennsylvania, Jan 2020 - May 2020 · 5 mos, Teaching Assistant for CIS 519 Applied Machine Learning, with Eric Eaton and Dinesh Jayaraman.

Teaching Assistant at University of Pennsylvania, Aug 2019 - Dec 2019 · 5 mos, Greater Philadelphia Area, Teaching Assistant of the graduate Machine Learning course CIS 520: Machine Learning, with Prof.Lyle Ungar.

Deep Learning Intern at Qualcomm, May 2019 - Aug 2019 · 4 mos, New York City Metropolitan Area

Research Intern at Indian Institute of Technology, Madras, Dec 2016 - Apr 2018 · 1 yr 5 mos, Greater Chennai Area, Worked with Prof.P.V.Manivannan in the Mechanical Engineering department at IIT Madras on probabilistic models for robotics. 1) Developed an algorithm based on the Extended Kalman Filter in a team of 2, which takes into account the correlations between measurement parameters. -Mathematically proved superiority to the original EKF. -Demonstrated validity of the algorithm in a Mobile-Robot Mapping problem by comparing it to the original EKF, which resulted in better estimates, lower uncertainties and faster convergence -Published the work in an international conference. 2) Developed a novel sensor-fusion algorithm to reduce measurement noise in SONARs, in a team of 2 - Used the gaussian correlation conjecture to find correlations between range and bearing errors - Combined Maximum likelihood estimates of SONAR measurements with pressure sensor measurements - Reduced range errors by 32% and bearing errors by 75% - Verified it experimentally - Published the work in an international conference 3) Implemented 2D SLAM on mobile-robots. - Used the ARIA C++ libraries to work with a simulation of AmigoBot. - Worked on EKF-SLAM with point features, using SONARs as range finders and odometry from wheel encoders - Implemented EKF-SLAM and Particle Filter based localization using ARIA and on MATLAB.

Research Fellow at Raman Research Institute, May 2017 - Jul 2017 · 3 mos, Bengaluru, Karnataka, India, Worked with Prof. Urbasi Sinha in the Quantum Information and Computing Lab at Raman Research Institute, as a part of the Research fellowship from the Indian Academy of Sciences Summer Research Fellowship programme. -- Studied about the loss of security in practice of the original BB84 Quantum Key Distribution (QKD) protocol due to Photon Number Splitting Attacks in multi-photon signal states -- Studied about Decoy State QKD as a solution to improve security, increase distance and rate of key distribution -- Described mathematical background and state-of-the-art advances in Decoy State QKD in a comprehensive final report

Education

Master's degree, Robotics, University of Pennsylvania

Bachelor of Engineering - BE, Electrical and Electronics Engineering, SSN college of engineering