💡 I like building things that live somewhere between software, hardware, data and a curious idea.
🔭 I am always looking for ways to improve things — or at least understand why they work the way they do.
🌱 These days I am learning how to find simpler solutions without making the problem smaller than it really is.
💬 Ask me about art, science, technology, strange prototypes or how to connect a sensor to something that was never meant to read it.
👨🏻💻 My first programming languages were C and Assembly. Python came later and quietly took over most of my work.
🧰 I enjoy owning the whole path: from a circuit, sensor or raw dataset to the interface another person actually uses.
🎓 Teaching is part of how I learn. I have led workshops and mentoring programs in Python, web development, IoT, electronics and creative technology.
🏗️ Philosophy: A model, device or prototype only becomes useful when somebody can run it, understand it and maintain it.
⚡ I like making fashion tech, creative coding experiments and art installations.
🎮 I am a terrible gamer, except when an N64 is involved. In Counter-Strike I answer to Melo.
🤸 I used to be a pretty good gymnast. My knee and a few other bones eventually submitted a different career proposal.
🧪 Many of my side projects begin with “I wonder if this could work…” and become a repository before I know it.
🚀 I work as a Machine Learning Engineer, building production data pipelines, engineering analytics and tools that turn complex outputs into useful information.
🗺️ I build end-to-end data products: ingestion, validation, modeling, APIs, deployment and visualization.
⚙️ I also work with embedded systems, Raspberry Pi, ESP32, sensors, communication protocols and hardware/software integration.
🐍 I care about Python code that is clean, modular, documented and boring enough to trust in production.
📝 I document architectures, interfaces, assumptions, tests and limitations because the next person should not need archaeology to understand a project.
🌐 I am interested in remote collaborations involving ML, data engineering, connected products, experimental interfaces and technical education.
A geospatial Machine Learning project that estimates residential property prices across France from almost one million historical transactions. It combines a LightGBM spatial model, temporal regression and hierarchical smoothing, then carries the result through PostGIS, FastAPI and an interactive Mapbox interface.
Python · LightGBM · pandas · GeoPandas · PostgreSQL/PostGIS · FastAPI · Docker · Mapbox
Open the live map →
A small browser laboratory for listening to what spatial processing actually changes. It synthesizes its own test sound and compares an original signal with orbital motion and binaural HRTF processing. No uploaded audio leaves the browser.
JavaScript · Web Audio API · HRTF · Real-time audio · Interaction design
Try the experiment →
A local-first encrypted vault prototype for sensitive records, beneficiaries and time-gated legacy sharing. The interesting part is not only the interface: the repository documents its threat model, trust boundaries and the reasons why a prototype should not pretend to be production-ready security software.
React · TypeScript · IndexedDB · Argon2id · XChaCha20-Poly1305 · Threat modeling
- Production ML & data products: feature engineering, validation, model integration and useful interfaces.
- Data pipelines: ETL/ELT, orchestration, reproducible processing and monitoring.
- Geospatial systems: PostGIS, spatial operations, geographic aggregation and interactive maps.
- Connected prototypes: Raspberry Pi, ESP32, sensors, local communication and cloud integration.
- Technical documentation: system architecture, interfaces, test evidence, limitations and hand-off material.
- Teaching & mentoring: Python, web development, electronics, IoT and project-based STEM learning.
