Computer vision & reinforcement learning researcher — building perception systems that see, track, and decide. Currently a graduate student at Northeastern University, working at the intersection of deep learning, real-time tracking, and autonomous control.
I work on the parts of vision systems that have to be right in real time — detection, tracking, and the geometry that ties pixels back to the physical world.
My research combines classical computer vision — feature descriptors, calibration, multi-view geometry — with modern deep learning pipelines for detection, tracking, and recognition. I care about systems that hold up outside benchmarks: latency budgets, occlusion, domain shift, the unglamorous failure modes.
On the RL side, I've trained PPO, SAC, and TD3 agents in the F1TENTH autonomous racing simulator, comparing them against classical control baselines. I'm drawn to problems where perception meets control — where a model has to look at the world and act on it before the next frame.
Outside the lab: Formula 1, Marvel marathons, and exploring Boston.
I'm open to research collaborations, summer internships, and full-time roles starting Spring 2027 in computer vision, perception, autonomy, and applied ML. The fastest way to reach me is email.
polagoni.m@northeastern.edu ↗