Research Interests

• Sequential Decision-Making Under Uncertainty: Developing MDP-based online control strategies using search heuristics, MPC, and stochastic sampling for cyber-physical systems.
• Reinforcement Learning & Multi-Agent Systems: Bridging RL (DDPG/PPO) with optimization (MILP) through policy shaping for safe, long-horizon decision-making.
• Vehicle-to-Building (V2B) Optimization: Designing negotiation frameworks and coordination policies that balance individual EV user goals with building energy management.
• Transportation Electrification: Grid-aware fleet optimization, transit simulation, and charging infrastructure planning for electric public transit.
• Cyber-Physical Systems (CPS): Building scalable digital twins and simulation toolchains for real-world energy-mobility coordination.

Technical Skills

• Languages: Python, C++, SQL, Java, JavaScript, MATLAB, LaTeX
• Optimization & Control: MPC, MDP formulations, MILP (CPLEX/Gurobi), RL (DDPG/PPO, RLlib), Search Heuristics, Game Theory
• Machine Learning: PyTorch, TensorFlow, Scikit-learn, Stochastic Optimization
• Simulation & Big Data: SUMO (Traffic Sim), GridLAB-D, PySpark, Digital Twins (OPTIMUS)
• Platforms & DevOps: Linux, Docker, Git, AWS (EC2/S3)
• Hardware & Maker: VLSI Design, 3D Printing, Drone/RC Plane Construction, Embedded Systems (Arduino, ARM), Microcontroller Programming, Sensor Integration
• Web & Visualization: D3.js, React

Projects

• Vehicle-to-Building (V2B) Optimization: Hierarchical control combining MILP and Multi-Agent RL to coordinate 50+ distributed energy assets; demonstrated >10% cost reduction with field data from Nissan.
• Electric Bus Scheduling & Charging Optimization: Grid-aware Stackelberg game for mixed-fleet scheduling in Chattanooga, TN; reduced costs by 14% and GHG emissions by 10 tons.
• MoveOD — Mobility Data Fusion Pipeline: Open-source pipeline fusing US Census and traffic data to generate calibrated Origin-Destination matrices. Won 2nd Place at INFORMS 2025.
• Emergency Response Center Planning: Spatial-optimization model using Nashville DOT crash data to identify optimal emergency hub locations.
• OPTIMUS Digital Twin: Discrete event simulator for energy-mobility coordination, enabling stress-testing under stochastic edge-case inputs.