I am an M.Sc. graduate from Sharif University of Technology specializing in robotics, reinforcement learning, and embedded machine learning. I am currently working as a Robotics Engineer at Fasta Robotics and actively seeking PhD positions in robust RL, adversarial training, and real-time embedded AI for robotic systems.
My research centers on robust control via zero-sum Markov games, where a learned adversary probes the weaknesses of a control policy during training, yielding formal worst-case performance guarantees. I work on dual-policy architectures that blend optimal and robust policies using transformer-based disturbance detection, and on deploying these systems in real time on NVIDIA Jetson platforms through TensorRT and ROS 2.
I completed my B.Sc. with the Best Undergraduate Thesis Award from the Iranian Aerospace Society for my work on controlling a 3-DOF quadcopter platform using Differential Game Theory and Nash Equilibrium.
Receiving the Best Undergraduate Thesis Award from the Iranian Aerospace Society.
My master’s thesis focused on zero-sum multi-agent reinforcement learning for robust spacecraft guidance, implementing DDPG, TD3, SAC, and PPO in adversarial game formulations with C++ real-time inference and ROS 2 hardware integration. I also contributed to projects involving neural-network-aided navigation (LSTM-based INS/GPS fusion) and differential game controllers deployed on embedded hardware.
Technical Skills
- RL & ML: PyTorch, JAX, deep RL (DDPG, TD3, SAC, PPO), multi-agent RL, adversarial training, transformer architectures
- Robotics: ROS 2, MuJoCo, Gymnasium, embedded control (STM32, Arduino), MATLAB/Simulink
- Deployment: ONNX, TensorRT, NVIDIA Jetson, CUDA, FP16/INT8 quantization
- Languages: Python, C/C++, MATLAB
I am actively looking for PhD opportunities to advance research in robust reinforcement learning and real-time robotic control.
Master thesis presentation (PDF)
My master thesis defense
