Research

Questions I'm pursuing, and space to pursue more.

I like research most when it has to survive real implementation: hardware that stalls, policies that lie to you, systems that fail in front of users. I'm always open to joining labs, collaborations, and projects where careful building is valued over polished claims.

Temporary stand-in photograph (the Eiffel Tower at dusk). Replace with a figure from the work.
Temporary stand-in. A hardware or lab still belongs here.

Three threads

  • Language environments

    Software that measures the speech children hear at home and collects real-time reports from their caregivers. LEAD Lab is the anchor.

    See LEAD Lab
  • Embodied systems

    Hardware bringup, cable-driven actuation, rover leveling control, and simulation built in parallel with the bench. SpiRob and VERTEX are the anchors.

    Open the case study
  • Learning systems

    Self-play reinforcement learning, search, and policy evaluation when rewards are sparse and every placement is permanent. BlokusDuo is the anchor.

    Read the paper

Language, Experience, and Development (LEAD) Lab · University of Maryland

Temporary stand-in photograph (a restaurant interior). Replace with a figure from the work.

Software for measuring a preschooler's day of language

SLIDE, a study of caregiver stress and preschool language. I built the survey-delivery system on the REDCap API and a Python package that computes speech burstiness from day-long LENA home recordings.

Recent work

  • Open the case study

    Study infrastructure and audio measures for preschool language development: real-time caregiver stress surveys and the burstiness of speech in day-long home recordings.

Role

Undergraduate research assistant · Dr. Rachel Romeo · Jun 2026 – present

Space Systems Laboratory - VERTEX rover · University of Maryland

Temporary stand-in photograph (a brick archway). Replace with a figure from the work.

Keeping a rover level on rough terrain

VERTEX, the Earth-analog chassis of the BioBot rover. I built its ROS 2 and Gazebo simulation and the leveling controller that keeps its deck within 1 degree of gravity on a 10-degree slope.

Recent work

  • Open the case study

    Simulation-first control for VERTEX, the Earth-analog chassis of the BioBot rover at the UMD Space Systems Laboratory: leveling, steering, and terrain sensing tested against known ground truth.

Role

Undergraduate researcher · 2025 – present

Zheng Research Lab · Drexel University

Temporary stand-in photograph (a flamingo chair). Replace with a figure from the work.

Staged bringup for a tendon-driven soft arm

Tendon-driven SpiRob soft manipulators in SolidWorks, staged cable control, and GPU-accelerated simulation in Isaac Gym and MuJoCo.

Recent work

  • Open the case study

    Experimental soft-robotics bringup: cable-actuated SpiRob, staged motor characterization, and parallel MuJoCo / Isaac Lab simulation.

Role

Research intern · Dr. Yue (Luna) Zheng · May 2025 – Jul 2025

Pennsylvania Governor's School for the Sciences · Carnegie Mellon University

Temporary stand-in photograph (an outdoor table). Replace with a figure from the work.

Self-play agents, and the paper that came with them

Residential program with a 7% admit rate. I led the BlokusDuo deep RL project and was main author on the peer-reviewed PGSS Journal paper.

Recent work

  • Read the paper

    Self-play reinforcement learning on a combinatorial board game. Minimax, MCTS, MCTS with a CNN, and PPO compared against random agents, each other, and human players.

Role

Research fellow and main author · Dr. Neil Simonetti · Summer 2024

  • Read the paper

    BlokusDuo deep reinforcement learning agents: architecture and experiments

  • Rare-cell recognition and quantum advantage

Papers and figures

If something here resonates, write to me.

If something here resonates, I'd like to hear what you're working on. I'm open to joining labs, collaborations, and projects where careful building matters more than polished claims.