Computer engineer working in applied AI: computer vision, sensor fusion, simulation, and automation. I turn complex data and system behavior into useful, reliable software.
Some work is summarized at a high level due to contractual and confidentiality constraints. No proprietary code, customer details, restricted datasets, or sensitive system information is included.
Sanitized Contract Work
Developed ML-based perception workflows using synthetic and real multi-spectral data to support object detection, sensor modeling, and fusion-oriented validation for technical R&D programs.
Sanitized Contract Work
Built AI-enabled internal tools that combine cloud services, vector search, LLM APIs, and automation workflows to accelerate technical analysis, documentation, and engineering decision support.
Sanitized Contract Work
Created simulation and modeling workflows for sensor behavior, environmental effects, engineering trade studies, and system-level R&D validation across defense and aerospace programs.
Implementation ability shown with open data, local tooling, and non-sensitive workflows across backend APIs, AI retrieval, and computer vision.
Project demo status
SentinelGrid, ProjectPulse, the Open Computer Vision Detection Dashboard, and the Technical Paper AI Search Assistant are all available as live public demos. Each repository includes setup or review instructions for exploring the project.
A local-first edge telemetry platform for climate-risk monitoring. It simulates 4,174 sensor nodes across 19 US regions, streaming readings through an MQTT pipeline into a geospatial store, with z-score + Isolation Forest anomaly scoring, sensor-drift quarantine, and incident tracking behind a national ops console. The dashboard overlays genuinely live public data (NEXRAD radar, ~3,700 real NWS/USGS stations, active storm warnings), and its deterministic engine can replay any moment of the last 24 hours, including storms that have already dissipated, without storing history. Runs fully in-browser, or against a live local backend.
What it demonstrates
A production-style project-management platform with an ASP.NET Core 8 Clean Architecture API and a React 19 dashboard. Features isolated demo workspaces, drag-and-drop Kanban board, file attachments, labels, role-based permissions, audit history, rate limiting, and 100+ automated tests across the stack.
What it demonstrates
A hosted computer vision dashboard where users upload an image, capture a webcam frame, or batch-process multiple images through live object detection. Features seven selectable models spanning three YOLO generations (v8–v12), the RT-DETR transformer detector, and open-vocabulary YOLO-World, which detects free-form text-prompted classes. Includes an interactive dark/light workspace with bounding-box viewer, original/annotated comparison, detection analytics, and annotated image, JSON, and CSV exports. Built with a Next.js/TypeScript frontend, Firebase Authentication, Storage, and Firestore for upload/job tracking, and a token-verified Cloud Run FastAPI inference service.
What it demonstrates
A source-grounded research assistant deployed on Cloudflare's free tier. Hybrid semantic + BM25 retrieval with reciprocal-rank fusion over a 10-paper aerospace corpus streams cited answers, highlights the exact passage on the real PDF page, and lets visitors search their own PDFs entirely in-browser, protected by a fail-closed daily quota so the demo can never generate a bill.
What it demonstrates
I'm Jeremy. I grew up in Hebron, Connecticut, where I went to RHAM High School and played football. A lot of who I am is still rooted in this corner of New England: hiking, swimming, and being out in the woods, plus a borderline unreasonable devotion to every Boston sports team and the UConn Huskies.
I studied at the University of Connecticut and was lucky enough to watch the Huskies go back to back, winning two national championships while I was there. That same mix of curiosity and competitiveness is what pulls me toward engineering. I like taking messy, real world problems apart and building software that actually holds up.
These days I work at the intersection of software, applied AI, and technical R&D, building with computer vision, sensor fusion, simulation, and automation to turn complex data and system behavior into tools people can rely on.
