Researcher · Robotics Engineer
Building robots that understand the world around them.
I work on autonomous systems, robot learning, and the messy gap between simulation and the real world. Currently finishing a degree and diving into the math behind why RL converges and optimal control works. Also — what do you think about large arrays of mirrors in an orbit beyond the GEO as a solution to daylight saving time?
Inked out
01
Semantic Observer Layer for Autonomous Vehicles
VLMs as real-time safety observers on top of standard autopilot stacks.
02
Symmetry-Aware Reward Shaping for Quadrupedal Locomotion
Teaching the Unitree Go2 to walk naturally using PPO + reward engineering.
03
Swarm-Based Debris Encirclement in LEO
Decentralized spacecraft formation control under orbital mechanics.
04
Decentralized Heterogeneous Swarm Flocking
Drones and ground vehicles coordinating through GNN-based goal assignment.
05
VLM-Driven Observer for Safety-Critical Maneuver Triggering
Cosmos-Reason1-7B as a closed-loop semantic observer inside Autoware on CARLA. 0 collisions vs. 7 at baseline.
Currently
- Humanoid playground — whole-body controllers + motion planners for grad students to explore
- Agile hexcopter — modelling in MuJoCo, fighting kinematic chain constraints
- VLM Observer paper — published, on-vehicle validation at NYU Brooklyn Commons is the next step
- Academic degree — finishing coursework, writing research