Selected systems · 2023–2026

Engineering at the edge of autonomy.

Aircraft, robots, and intelligent systems built to perceive, decide, and act in the physical world.

AI Grand Prix ↗ · Team Lead · 2026

Leading Georgia Tech at the AI Grand Prix

As team lead, directed an eight-person MS/PhD team building Georgia Tech's complete autonomous drone-racing stack, from perception and visual-inertial odometry through planning and nonlinear control.

Virtual qualifier finish: 3rdout of 3,000+ teams worldwide
Overall course time: 13.52seconds to finish

Live simulation

ATTITUDE−12.4° ALTITUDE4.8 m

OVERALL00:13.520 LOAD6.27 g

Role: Team Lead
Team: 8
Max speed: 113.3 km/h
Average: 53 km/h
Peak load: 6.27 g

Predictive flight controls · Real time

Seeing turbulence before the aircraft reaches it

Designed and implemented a predictive suppression system that models the LiDAR and aircraft dynamics, estimates the incoming disturbance, and solves an MPC counter-response in under 100 milliseconds.

Incoming disturbance Atmospheric turbulence

Look-ahead sensing Doppler LiDAR

Counter-response Aircraft stabilized

Predicted disturbance envelope

01 · Sense Measure airflow ahead Forward-looking Doppler returns capture the disturbance before encounter.
02 · Estimate Reconstruct the disturbance Sensor and plant models turn raw velocity data into a control-ready state.
03 · Optimize Solve the MPC response The controller predicts aircraft motion and selects the best counter-command.
04 · Act Reject before impact Flight controls move the aircraft against the incoming load in real time.

Project outcome: the work led to

Airline letters of intent: $14Min LOIs
Seed financing: $750Kraised

My contribution

From sensor physics to flight code.

Architecture

Designed the full LiDAR-to-actuator control pipeline.

Controls

Built the real-time model predictive controller.

Software

Implemented the stack across Python, C++, and Rust.

Validation

Developed plant and sensor models in MATLAB/Simulink.

End-to-end latency <100 ms

Core method Doppler LiDAR + MPC

Engineering context Avionics · Certification

Research · Controls · Autonomous flight

Learning and optimization for autonomous flight

Designed an eVTOL autonomy stack for takeoff, navigation, and precision landing, comparing classical control with sampling-based and learning-enabled methods under uncertainty.

Trajectory comparison

Position error / time

LQR
MPC
MPPI
Reference

Also in motion

More systems,
same obsession.

Bipedal locomotion

Hybrid reinforcement learning and batched MPC for robust, terrain-adaptive locomotion and Sim2Real control.

Robotics · RL · MPC

Hyperspectral plant intelligence

Vision-language models, CNN/U-Net segmentation, and active learning for plant-health prediction across changing crops and light.

75–80% accuracy · VLM · Edge AI

Event-driven FX engine

A low-latency research architecture for ingesting market events, evaluating signals, and coordinating execution.

C++ · Rust · Python

GT Supersonics

Founded the team and led propulsion testing, performance validation, and gas-turbine integration for a supersonic RC aircraft.

Propulsion · Flight test · Leadership