Projects
Deployed Systems
Lidar Fusion Project
Penn State Advanced Vehicular Team
Real-time 3D spatial localization using Sensor Fusion
Engineered a real-time sensor fusion pipeline in ROS 2 Humble using Python and OpenCV, implementing robust synchronization of high-frequency LiDAR and camera data via Best Effort QoS policies. Integrated this fused output with YOLO object detection to assign precise distance metrics to 2D bounding boxes, enabling accurate 3D spatial localization for dynamic obstacles.
Personal Portfolio V2.2
Immersive 3D/WebGL Identity Platform
A high-performance site built on Next.js 15. Features a custom 'Silk' WebGL fluid sim via R3F with GPU shaders, glassmorphism UI with Framer Motion animations, and a fully typed, edge-deployed architecture on Vercel.
ChatDnD
Multi-Agent AI Dungeon Master with LangGraph
Built a conversational AI system using LangGraph and MongoDB for managing complex multi-agent interactions in D&D gameplay. Integrated WebSockets for real-time communication.
Bayesian Portfolio Optimization
LSTM stock forecasting with Bayesian Networks
Developed a sophisticated portfolio management system combining LSTM neural networks with Bayesian inference for stock price prediction using Alpha Vantage data.
Distributed Storage System
Linearized file system in C/C++ with multithreading
Engineered a high-performance distributed file system with advanced concurrency control and linearizability guarantees.
In the Pipeline
BuxBalance🚧Dev
High-Fidelity Financial Command Center
A hybrid personal finance platform combining the aesthetic visibility of Monarch Money with the subscription rigor of Rocket Money. Features a 'Magic Bento' UI, AI-powered transaction categorization using Gemini 2.5 Flash, and a massive interactive Sankey diagram for cash flow visualization.
BrickByBrick (B3)🚧Dev
Supply Chain Optimization Engine for AFOLs
An advanced sourcing engine decoupling the 'Part' from the 'Box.' It utilizes a specialized algorithm to source parts for retired Lego sets at a fraction of the cost. Features include a 3D R3F visualization of sets, occlusion analysis for internal parts, and a multi-mode sourcing engine (Purist vs. Builder modes).
FireWatch🚧Dev
WRF-SFire Wildfire Simulation (HPC)
A supercomputing initiative running on the Penn State ICDS Roar Cluster. This project compiles and executes the WRF-SFire model to simulate the 2018 Camp Fire in Butte County, CA. It involves complex compilation of Fortran/C static libraries to resolve architecture-specific (SIGILL) errors and integrates high-resolution LANDFIRE fuel data.