Developing novel machine learning and reinforcement learning algorithms that leverage large language models as a natural interface between humans and data-driven systems. Specifically, I’m interested in how these algorithms can be used to improve communication and decision-making in scenarios where humans and machines need to work together.

Building software tools to make individual traders more confident when investing their money.

• Fused human eye gaze and action demonstration data using a multi-objective imitation learning architecture to improve performance by 14% on a simulated autonomous quadrotor search and navigation task when compared to single-objective imitation baselines.• Integrated automatic curriculum learning generation to deep reinforcement learning algorithms using a single human demonstration to solve multi-agent learning scenarios, outperforming both imitation and reinforcement learning baselines.

• Researched novel deep reinforcement learning and supervised learning algorithms to develop autonomous agents, computer-vision algorithms for sensor fusion and object detection, and data-driven methods to control autonomous systems.• Developed an algorithm to utilize different forms of human interaction to safely train autonomous systems in real-time, increasing task performance by 13% while requiring 33% fewer data samples, while also improving by itself via reinforcement learning.• Authored research proposal on human-in-the-loop learning awarded funding ($750K/3yr) by the US Army Research Laboratory.• Received US Army Research Lab’s Award of Science for human-in-the-loop learning research, 2019.

• Developed novel algorithm to utilize different forms of human interaction to safely train autonomous systems in real-time, increasing average task completion by 13% while requiring 33% fewer data samples.• Published research results at the 2019 AAAI main conference and at the 2018 AAAI Symposium.• Won A-Hack-of-the-Drones, vision-based drone detection, sponsored by MD5/Army Futures Comm., 2018.

Development of computer vision-based behavior cloning and human-in-the-loop reinforcement learning algorithms for small unmanned air systems applications. This work was done during the Summer Journeyman Fellow in the U.S. Army Research Laboratory (USARL) Research Associate Program, Sensors and Electron Devices Directorate (SEDD), supervised by Dr. William Nothwang. This work was published at the 2018 WCCI conference with the title "Cyber-Human Approach For Learning Human Intention And Shape Robotic Behavior Based On Task Demonstration".

Developed the LASR Attitude Test-bed (LAT), an open-source spacecraft attitude estimation and control test-bed, a low-cost sun sensor for spacecraft attitude determination, and supported the development of HOMER II - Holonomic Omni-directional Motion Emulation Robot.

Developed studies to expand and design new power substations in 34.5 and 138 kV to attend the power demand of new customers. This included studies for the better connection arrangement in order to minimize the construction costs and planning for future expansions.

Developed an impact study of multiple solar plants connected to the grid and how it could benefit or damage the grid if precautions were not implemented on the grid’s protection and flow control system.

Developed studies of the impact of grid-connected photovoltaic microgeneration to the company’s grid, which later led to the development of my Bachelor’s thesis. I worked with industry partners to develop standards for residential installation of photovoltaic systems and connection to the utility grid resulting on the company’s first set of procedures allowing customers to install these systems and sell the generated energy to the grid.

Part of the team who developed a novel Tri-State Boost Current Source Inverter (CSI) for a water pump powered by solar panels. This research was presented on the 2011 IEEE International Future Energy Challenge Competition (IFEC’11) Competition, awarded the IES Best Innovative Design of Power Electronic Converters.