Hello,
I'm Valerie.

Studying Computer Science, Applied Mathematics & Statistics, and Cognitive Science at Johns Hopkins University.

Passionate about machine learning, robotics, and computer vision — exploring how algorithms can perceive, reason, and interact with the world.

Languages

PythonC++Java CJS/TSMATLABSQL

Interests

ML / AIComputer Vision Autonomous SystemsRobotics

Python · PyTorch · NLP

Causal Intervention Analysis of Coconut LLM

Steering vector experiments for Coconut LLM. Extract and analyze steering vectors using contrastive pair methods (swap / cross-pair) to manipulate latent reasoning trajectories and evaluate superposition effects on question answering.

Python · ROS2 · ZED SDK

JHU URC Rover Autonomy Pipeline

Real-time autonomous navigation pipeline for the JHU Mars Rover Team's URC competition entry. Fuses ZED stereo visual odometry, scene understanding (wall/hallway detection, depth clustering), and YOLOv8 object detection into a unified loop.

SQL · Python · PHP

Asteroid Risk Database

Processes 40,000+ NASA JPL Near-Earth Objects through a Random Forest classifier (AUC > 0.99) and anomaly detection to generate risk scores (0–100) for every NEO. Interactive web interface with 16 analytical queries, orbital visualizations, and real-time threat ranking.

Python · SLAM · EKF

Monte Carlo EKF-SLAM Simulator

Active EKF-SLAM simulator for autonomous exploration of unknown 2D environments. The robot constructs a feature-based map using an Extended Kalman Filter and selects navigation goals through Monte Carlo sampling driven by positional uncertainty and information gain.

Python · RL · PDE Systems

Fluidic Pinball Control

Reinforcement learning for closed-loop control of multi-physics fluid systems at Lawrence Berkeley National Lab. Frames problem as POMDP with actor-critic methods on spatio-temporal Fourier transformer.

JavaScript · Chrome Extensions API · HTML/CSS

Video Speed Controller

Universal Chrome extension for HTML5 video speed control across any site. Features persistent speed per domain and Picture-in-Picture.

See more projects ↗

2026

Decentralized multi-agent reinforcement learning for hybrid truck-drone last-mile delivery.

Liang V., Deniz, S., Chakraborty, S.

Transportation Research Board 105th Annual Meeting, 2026

Abstract ↗