Software and data engineer, open to contract work.
I build AI products, data systems, and web applications end to end, from the first conversation to the infrastructure underneath.
I’ve built software at Trimble and Novartis, and data infrastructure for the Center for Food Action (Ramapo College DMC). Most of my work is open source, and I’m always building something.
Adding AI to a product or workflow you already have, or building a new one around it. Model integration, tool calling, retrieval, memory, and agents with clear boundaries, so a person still reviews the steps that matter. I also set up evaluation, so you can tell whether it’s actually working.
LLM APIs · Tool calling · Agents · RAG · Evaluation
Moving data from where it’s collected to where people use it. Ingestion, cleaning, validation, storage, and reporting, with review steps where accuracy matters. Recurring manual reports become dashboards that refresh on their own.
Python · SQL · PostgreSQL · ETL · Power BI
From the first idea through the interface, the backend, and deployment. That includes search engine optimization, performance, and accessibility compliance, so the site is easy to find and works for everyone. Security is part of the build: authentication, role-based access, and keeping your data protected.
React · Next.js · Node.js · PostgreSQL · SEO · WCAG · AWS
Tools that bring scattered systems, settings, and permissions into one workflow. Connecting the APIs you already rely on, automating the handoffs between them, and setting up access control so the right people see the right things.
C# · .NET · Azure · REST APIs · Automation
Organizing your data so people can find what they need. Search across documents and records, semantic search, knowledge graphs, and ranking that reflects what someone asked for, not just the keywords they typed.
Embeddings · pgvector · Knowledge graphs · Ranking
Most of what I build is public.
A graph-aware retrieval library for intent-content conflict. Bring your own embeddings and knowledge graph; it trains a GNN and scores results on content and intent.
Python · GNNs · Knowledge graphs
The research code behind my NGEN-AI 2026 paper: graph-grounded retrieval for intent-content conflict, evaluated on affective movie retrieval.
Python · Jupyter · MovieLens 25M
A film diary and recommendation system built on what films meant to people, with the person behind each recommendation still visible.
React · TypeScript · Express · PGlite
A local agent workflow system with Hermes tools, structured state, provenance, and human approval for external steps.
Python · SQLite · Hermes Agent · Typst
Context for in-person recruiting: candidate-written meeting notes, shared work, and source-linked role lookups.
React · TypeScript · Node.js · SQLite · Rust
An AI study partner over a student’s Canvas data, with persistent memory, tool calling, and semantic search. Won HackRamapo.
Next.js · PostgreSQL · pgvector
A shared room-layout canvas for people and browser agents, built on WebMCP.
TypeScript · WebMCP
A lean local knowledge base with SQLite, CLI search, and structured distillation, so useful context stays searchable without another cloud dashboard.
Python · SQLite