About
I’m an AI and robotics research engineer working at the intersection of multimodal intelligence, 3D vision, and autonomous systems. Over the years, I’ve built localization and planning systems for drones, monitoring frameworks for self-driving vehicles, reinforcement-learning planners for mobile robots, and reconstruction pipelines for challenging 3D scenes. I’m most interested in turning ambitious research ideas into systems that can perceive, reason, and act reliably in the real world.
I’m exploring faster, more composable simulation tools that make it easier to build realistic environments and vary scenarios at scale. The goal is to shorten the path from an idea to a reliable experiment—and eventually to a policy that works on a real robot.
I’m studying how flow-based generative models can learn coherent action sequences and represent multiple plausible futures. I’m especially interested in world action models that connect learned dynamics with planning and closed-loop robotic control.
Experience
Across foundation models, 3D reconstruction, autonomous vehicles, and aerial robotics.
Meta
Multimodal AI research.
Quidient
Led reconstruction work for highly specular objects, dense point clouds, and sub-pixel camera alignment.
Nuro
Built a production-ready, hardware-agnostic framework for monitoring autonomous-vehicle actuators and controllers.
Maryland Robotics Center
Developed an indoor delivery robot and reinforcement-learning planners for Nav2.
FlytBase
Built localization, planning, and safety systems for commercial autonomous drones.
IIT Bombay
Worked on dynamic path planning and robust control for indoor micro aerial vehicles.
Open Source
A selection of open-source explorations across neural rendering, reinforcement learning, reconstruction, and robotics.
A playful lenticular effect built from a neural radiance field reconstruction.
An elegant implementation of neural graphics primitives for 2D and 3D reconstruction.
AlphaZero and Monte Carlo tree search applied to Tic-Tac-Toe and Connect Four.
An autonomous mobile manipulator that finds, picks, and organizes misplaced household objects.
A compact implementation of neural radiance fields for novel-view synthesis.
Live RGB-D fusion using a truncated signed distance function.
LQR, LQG, and observer-based control for stable payload motion.
Camera motion estimation built from geometric constraints and image correspondences.
Smooth, constraint-aware robot trajectories represented with splines.
A mobile robot integrating manipulation, kinematics, and autonomous control.
Toolkit
Contact
Whether you’re exploring a research question, building an ambitious product, or finding your way into the field, I’d be glad to hear from you.
Let’s collaborate on ideas across multimodal AI, robotics, simulation, world models, and embodied intelligence.
I’m open to focused consulting on AI and robotics strategy, technical direction, and turning research into practical systems.
If you’re navigating robotics, AI research, graduate school, or an early career decision, I’m happy to share what I’ve learned.