Perception Engineer · USA
I build the perception and autonomy behind intelligent machines.
Founding engineer at Space Intel AI, where I lead perception and autonomy for aerial platforms. I’m interested in deep technical systems that combine machine intelligence with human judgment to solve real-world problems. I like tackling hard, ambiguous problems and reducing them into systems that are technically sound, observable, and reliable.
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
StratoShield
Geofencing critical infrastructure from high-altitude aerial surveillance. I own it end to end, from product direction to the perception itself: detection, tracking, localization and mapping, and a synthetic-data pipeline that lets us develop before the hardware exists.
Depth from Sound
PeAR Group, WPI ↗Estimating dense depth maps from reflected audio with a temporal convolutional network, teaching a model to read a space from the way sound comes back to it. I built the acoustic raytracing simulator behind it in Blender and Python to generate the training data.
Humanoid Tele-nursing Robot
HiRO Lab, WPI ↗Research on making tele-manipulation easier to operate. I built shared and autonomous camera-in-hand viewpoint control for manipulation tasks with 7-DOF Kinova arms (on real hardware and in a Unity hospital simulation), and AR cues for assisting mobile teleoperation and tele-manipulation, evaluated through user studies with gaze tracking.
Autonomous Underwater Vehicle
DTU-AUV Lab ↗Years at the DTU-AUV Lab building AUVs and ROVs: mechanical design, vehicle modelling, controls, and perception, for tasks like object picking, dropping, and torpedo launching. The vehicles held their own at national and international competitions.
Autonomous Surface Vehicle
An autonomous boat that collects floating waste from urban drains, a problem posed by India's Department of Water Resources. I worked on the object detection, motion planning, and controls. It went on to win the Smart India Hackathon.