B.Sc. in Robotics & Mechatronics Engineering
Fourth year (8th of 8 semesters), ongoing. CGPA 3.50 / 4.00. Expected August 2026.
A summary of my background, research, and training. For the full record, see my CV (PDF).
I am a final-year B.Sc. student in Robotics & Mechatronics Engineering at the University of Dhaka, working on reinforcement learning for locomotion, optimization-based control of dynamical systems, and Sim2Real robot learning. As an undergraduate researcher in the MARVEL Lab, I am the lead author on deep RL controllers that recover rover mobility in deformable Mars regolith, learning locomotion policies where classical control fails under granular slip.
My interests sit at the meeting point of safe and preference-driven reinforcement learning, learning-based control, and neuro-robotics for movement. Alongside research, I build real systems, from STM32 flight controllers to ROS2 navigation stacks, and I care about closing the gap between what a policy learns in simulation and how it behaves on hardware.
Fourth year (8th of 8 semesters), ongoing. CGPA 3.50 / 4.00. Expected August 2026.
Sylhet, Bangladesh. Board rank 62 (Sylhet Board, 2021).
Lead author on DRL controllers that recover rover mobility in deformable Mars regolith. Designed yaw-projected reward shaping with entrapment-gated slip penalties (a safety-style constraint) and built large-scale, multi-GPU training and evaluation pipelines.
Building a digital-twin warehouse benchmark to evaluate Deep RL navigation policies (PPO, SAC, TD3) under realistic perception and localization noise.
Studied melanoma segmentation from pseudo-annotations under limited labels (weakly-supervised computer vision).
STM32-based autonomous drone platforms for maritime monitoring, robust firmware, hardened telemetry, weather-resistant sensing.
Developing autonomous robots for industrial and educational use with ROS2 navigation and perception stacks.
Proposed the concept at IEEE SB DU; leading design, frame integration, and fundraising for a competition-ready rover platform.
Lead chapter activities, events, and research collaborations promoting robotics and automation among students.