🎓 M.Eng. in AI & Robotics | 🤖 AI/ML Developer | 🛠️ Problem Solver
Welcome to my GitHub! I'm passionate about building reliable and efficient AI/Robotics systems and applying machine learning to solve real-world problems.

I'm a Master of Engineering (M.Eng.) graduate from the University of Toronto (MIE Department), with dual emphases in Robotics and Data Analytics & Machine Learning. My experience focuses on applying ML in various practical engineering settings — from robotic localization to optimization.

I enjoy turning complex problems into working, maintainable software and am seeking full-time opportunities in ML engineering, robotics, or AI applications.

• Programming: Python, MATLAB, C++ (basic), SQL
• Libraries/Frameworks: PyTorch, ROS, OpenCV, NumPy, scikit-learn
• Domains: Robotics, CV/ML, Dataset Distillation, Reinforcement Learning
• Tools: Git, Jupyter, Linux, VS Code, Azure

• Franka Golf — RL Competition Solution
  Curriculum-based reinforcement learning (SAC + reward shaping) for a Franka Emika Panda robot to play a golf task.
  Ranked 6th out of 59 competitors in the CompeteSAI challenge.

• Autonomous House-Organizing Robot (repo in progress)
  Hybrid navigation system combining classical planning and DDPG policy for autonomous exploration in simulated indoor environments.

• Robot Data Collection Framework Using ROS and Gazebo
  Built a ROS framework for simulating peg insertion and object manipulation with Cartesian compliance control.

• Hybrid Optimization & RL for EV Routing
  Developed a hybrid reinforcement learning and optimization framework for electric vehicle routing with energy and time window constraints.

• Privacy-Preserving Synthetic Data Generation
  Designed a privacy-preserving data generation pipeline using a modified variational autoencoder for tabular datasets.

• Dataset Distillation for Efficient ML Training
  Reproduced and analyzed the DataDAM and PAD method for efficient dataset distillation using attention mechanisms.

• State Space & Vision-Language Models (repo in progress)
  Explored scalable architectures like Mamba and proposed optimizations to reduce computational bottlenecks in vision-language models.

• CFD Solver from Scratch (C++)
  Implemented a 2D Navier-Stokes solver to simulate fluid flow between plates and smoke dispersion.

• Cooling System for IoT Gateway
  Designed thermal control strategies and ran finite element simulations for passive cooling in compact IoT enclosures.

• Human Motion Modeling in Simulink
  Simulated biomechanical movement for physical therapy, using inverse and forward dynamics of triceps dip motion.

• Object Detection for Recycling Bins (Arduino)
  Built and deployed a portable ultrasonic sensor system to detect nearby objects around public recycling bins.

• Motion planning (PRM, RRT, RRT*)
• Wheel odometry and grid-based mapping
• Lidar mapping and localization
• Particle filter localization
• Estimation methods: batch linear-Gaussian, EKF, Gauss-Newton

• Linear models, convolutional models, generative models, sequential models
• Comparative study on text classification models (CNN, SVM, RNN)

• K-Means clustering using MapReduce
• Recommender systems using Spark
• SQL orchestration and stream processing in cloud environments
• Information retrieval and prompt engineering

• Cash Matching Portfolio Optimization
• Efficient Frontier Construction

• 🔗 LinkedIn
• ✉️ Email: pas.panthasen@gmail.com