Technology

The technology I design and build.

A classical marble head dissolving into cubic fragments and exposed circuits amid dusty pink and blue smoke

From data to software.

I work on AI engineering, software development, data science and technical advisory. In practice, I build applications, put models and agents to the test, and help teams make the technical calls that shape a product.

My GitHub

How I work with technology.

  • AI engineering

    I put AI models to work inside real tools and workflows: agents, document retrieval, computer vision. I always start from what the model can and can’t do in that context.

  • Software development

    I build applications and tools, from the backend to the interface. I care about architecture, integrations and code quality, and about what keeps software alive over time: performance, tests, maintainability.

  • Data science

    I work with data from cleaning and analysis to predictive models. A model is only useful once you’ve measured it honestly, understood where it fails and wired it into the software.

  • Technical advisory

    I help teams and companies with technical decisions: comparing approaches, choosing an architecture, deciding whether to build or buy. The point is to match what technology can do with what the product actually needs.

Close-up of fractured marble in dusty pink, with dark circuitry exposed and stone fragments drifting into smoke blue

Projects, with the code.

A selection of public projects: tools, applications and experiments that show different sides of my technical work.

  • MAGI System

    Python / Document retrieval / AI agents

    A command-line assistant for querying your own documents. It retrieves relevant evidence, compares three AI reviewer perspectives and returns an answer with citations.

  • Chronos

    Rust / Async runtime / Debugging

    A deterministic async runtime in Rust, with a virtual clock and a terminal debugger. It records and replays execution traces to investigate concurrency bugs; external I/O needs separate handling.

  • MarkGuardiola

    Python / React / FastAPI / Machine learning

    A local fantasy-football assistant. It brings together data pipelines, predictive models and a web application to support lineup and transfer decisions.

  • Computer Vision TR-808

    Python / OpenCV / MediaPipe / Streamlit

    A virtual drum machine that turns finger movements captured by a webcam into TR-808 sounds. An experiment combining computer vision, interaction and music.

Let's talk.

Tell me whatever comes to mind.

20–5,000 characters. All fields are required.

I use your details to reply to your message, as described in the privacy notice.