Perception Engineer · USA
I build perception and state‑estimation systems for robots.
Founding engineer at Space Intel AI, where I lead computer vision for autonomous aerial and stratospheric platforms. I like geometry done carefully, filters that stay consistent, and tools that show you what the robot sees.
Featured work
msckf-vio
A from-scratch Multi-State Constraint Kalman Filter for visual-inertial odometry, in NumPy — verified in simulation, then made to survive real flight data.
- ATE RMSE on EuRoC MH_01
- 0.23 m
- drift over 80.4 m of flight
- 0.29 %
- in pure Python
- ~70 fps
sfm-core
Incremental structure-from-motion written from first principles — every estimator hand-built in NumPy, unit-tested against synthetic ground truth.
- mean reprojection error
- 1.6 px
- 3D points from 5 images
- ~1,000
- full pipeline on a laptop
- < 4 s
bev-stack
ActiveA bird's-eye-view perception viewer built around one canonical scene contract — backends adapt datasets and models, the renderer only ever sees SceneFrames.
Read the case study →Also on the bench
slam-core
ActiveA geometry-first visual-inertial SLAM system in modern C++, built in disciplined stages with Rerun-based debugging and reproducible EuRoC experiments.
frame-lab
ActiveA LeetCode-style practice lab for perception algorithms — small, testable Python exercises for the geometry and detection math everyone uses and few implement.
Beyond the repos
Depth from sound
Depth mapping from reflected audio using a TCN, plus an acoustic raytracing simulator in Blender that generated 41k synthetic images. Won the Dr. Glenn Yee Project Award at WPI.
StratoShield
Designed the perception stack — detection, tracking, sensor fusion — for geofencing across drone and stratospheric platforms, and picked the full sensor suite under strict SWaP constraints.
An autonomous underwater vehicle
Perception and sensor integration for AUV VARUNA 2.0 at DTU — a finalist at the Singapore AUV Challenge, now displayed at the National Robotics Museum of India.
A trash-collecting boat
An autonomous surface vehicle with a ROS mission planner and real-time detection on a custom dataset. Won the Smart India Hackathon.