Work

FleetOps AI combines retrieval-augmented generation, LLM orchestration, and human oversight to support industrial incident analysis. Designed with a focus on reliable backend engineering, traceable AI outputs, and cloud infrastructure.

Built a custom Python constraint-satisfaction engine with 26 solver configurations spanning depth-first search, constraint propagation, and arc consistency. A FastAPI backend and React/TypeScript interface enable live algorithm inspection, historical replay, and controlled performance experiments.

Built a full-stack TypeScript application that transforms PDF collections into searchable knowledge sources. A NestJS backend manages document processing, vector retrieval, and conversational AI, while a Next.js interface provides streamed answers with direct navigation to supporting PDF pages.

Designed and implemented an aircraft monitoring system built around a C++20 networking backend and a React/TypeScript operator console. The system processes aircraft position updates, maintains persistent flight sessions, and provides live tracking, geographic boundary detection, and historical analysis.

Designed and implemented a modular voice communication system focused on signaling, media delivery, and network resilience. The architecture separates call control from asynchronous packet transport, with reproducible failure scenarios and live monitoring of system behavior.

Redesigned a computationally expensive 3D CT pattern-matching algorithm using compact 64-bit encoding, stable radix sorting, and linear aggregation. Implemented and benchmarked sequential, POSIX Threads, and MPI versions, achieving up to 3.95× acceleration over the optimized sequential baseline.

Developed two interconnected Unity applications for immersive robotic control and real-time simulation. A Meta Quest 3 interface translates hand-tracked interactions into ROS 2 joint commands, while a desktop application applies the received positions to an articulated xArm 1S model.

Unity/C# system enabling therapists and patients to perform guided upper-limb rehabilitation exercises in a shared virtual environment. Implements OpenXR and Meta XR interaction rigs, real-time hand and finger synchronization, hybrid hand/controller input and configurable avatar scaling. Tested on Meta Quest 2 and 3.

Python desktop application that processes structured patient data, runs a trained TensorFlow/Keras neural network and stores measurement results locally for later review. The system combines data preprocessing, offline inference and a CustomTkinter interface in a complete end-to-end machine learning workflow.

Developed three implementations for identifying matching 4 × 4 × 4 voxel blocks in volumetric CT data. The system uses compact 64-bit pattern encoding and distributes exhaustive pairwise comparisons across POSIX Threads and MPI processes to explore different parallel execution models.

Designed and implemented a full-stack Retrieval-Augmented Generation application for querying PDF documents through natural language. The system integrates a NestJS backend, Supabase vector storage, LangChain orchestration, and a Next.js interface for document management and question answering.

Built a Python ETL pipeline that retrieves earthquake data from the USGS feed, processes geographic and temporal attributes, and persists records in SQLite. A FastAPI backend provides magnitude-based filtering, daily statistical summaries, and tsunami-alert analysis through documented endpoints.

Developed an interactive Java Swing game combining player-trained automation with progressive gameplay mechanics. The application measures manual click intervals to control automated scoring, integrating vehicle upgrades, fuel management, speed boosts, achievements, and custom animations.