cv

MS Robotics

University of Pennsylvania,Philadelphia, United States

• Thesis - Learning goal-conditioned control policies for waypoint reaching in aerial vehicles and aerial manipulators (in progress)
• Coursework - Learning in Robotics, Advanced Computer Vision, Machine Learning Theory, Advanced Robotics, Model Predictive Control, and F1-Tenth Autonomous Racing
• Advisors - Dr. Vijay Kumar and Dr. Pratik Chaudhari

B.Eng, Mechanical Engineering + Minor in Computer Science

National University of Singapore, Singapore, Singapore

• Advisor - Dr Guillaume Sartoretti and Dr. Ang Marcelo H Jr.
• Coursework - Artificial Intelligence, Machine Vision, Advanced Robotics, Neural Networks, and Linear Systems Theory

Research Assistant

Kumar Lab at GRASP, University of Pennsylvania, Philadelphia, US

• Project 1
  • Developed a learning-based motion planning method with motion primitives to localize targets via quadrotors.
  • Trained and benchmarked light-weight transformers and RESNET networks with RL algorithms such as PPO, SAC and A3C and benchmarked performance with greedy heuristics, CMA-ES, MIPP and MCTS
  • Demonstrated learnt planner on Kumar Autonomy Stack with Gazebo Simulations
• Project 2
  • Adapting multi-agent reinforcement learning for cooperative and persistent target search and tracking
• Project 3
  • Researching methods for learning goal conditioned control policies for referenctracking and waypoint reaching for aerial vehicles and aerial manipulators that can be leveraged for general aerial vehicle tasks

Teaching Assistant

University of Pennsylvania, Philadelphia, Pennsylvania, US

• Teaching Assistant for CIS 521 - Introduction to AI, CIS 522- Introduction to Deep Learning, ESE 546 - Principles of Deep Learing and CIS 519 - Machine Learning at UPenn.
• Responsible for conducting weekly office hours and recitations, grading homework and content modification.

Research Engineer

National University of Singapore, Singapore

• Introduced adaptions to a multi-agent RL algorithm COMA for fully decentralized and stable cooperation learning in large scale traffic systems
• Demonstrated that this led to performance improvements of over 20% in terms of traffic throughput in large scale urban networks comprising over 200 traffic signals
• Developed Graph Neural Networks and attention-based models for cooperative decision-making in traffic signal control

Robotics Software Developer

Movel AI, Singapore

• Adapted RRT based motion planning algorithms for site inspection with MovelAI’s software stack on ROS
• Developed UI on node.js with real time mapping and visualisation of faults during robot operation

Undergraduate Researcher

National University of Singapore, Singapore

• Researched evolutionary optimization approaches for information gathering for mapping with individual robots
• Developed multi-agent simulation environment in Gazebo to benchmark multi-agent mapping algorithms

Research Development Engineer

Thales Research and Technology, Singapore

• Investigated and formulated optimal control algorithms for coverage control via multiple satellites
• Tested robustness of control algorithms to state estimation and sensor fusion via Kalman Filters (UKF’S and EKF’s)
• Optimised and tested control algorithms for deployment on real satellites in Low Earth Orbit on Thales satellite simulator

• Dean's List

• IEEE NUS Hackathon Winner

• NUS N-House Pitch Night Winner

• Dean's List

• Rank 104 in JEE Main in IIT
• Rank 1387 in JEE Advanced in IIT