Applied researcher — computer vision, remote sensing & industrial process automation.
- Years in industry & applied R&D
- 14+
- Patents
- 2
- Applied R&D projects
- 13
Research & R&D
Curriculum vitæ
Contact
Selected R&D projects
A local-first agent stack that turns heterogeneous personal data into a provenance-tracked knowledge graph, queried by a sandboxed local LLM through read-only tools, with every interpretation validated by a human.
An unattended, resumable Windows-native pipeline that turns handheld 4K video into dense point clouds, textured meshes, orthographic views, Gaussian Splatting scenes and web exports on a shared RTX 5090.
A reproducible, provenance-tracked spatial warehouse that reconstructs who is in Fort-de-France city centre, where and when, from open data over 2016–2026.
A Python library that consolidates production document-processing systems into small, typed, traceable atoms, assembled into validated pipelines that return tri-state, explainable verdicts.
A print-and-scan data channel that stores 2–10 MB media files on ordinary A4 paper at about 383 KB per page, recovers them byte for byte, and prints its own decoder.
A provenance-first method for turning field-observation deliverables and legacy business software into typed, sourced knowledge that an AI assistant can query without confusing evidence with generalisation.
A local, rule-based robot that turns daily alarm reports, a web portal without public API and client spreadsheets into reviewed Outlook drafts for four recurring processes — with no language model in the loop.
An offline-first system that matches a circulating vehicle fleet against a manufacturer recall list, coordinates mobile field units and tracks each identified vehicle through to completed repair.
A pipeline that turns multitrack conference recordings into a speaker-labelled transcript and a pre-cut AAF timeline that opens natively in professional audio editors.
A monocular pipeline that turns fixed-camera and aerial video into GPS-referenced object detections, tracks and metric distances, using YOLO11 and Depth Pro at native resolution.
Drone-based detection of gas and hydrocarbon pollutants, operational day and night, for maritime and industrial zones — fuel leaks, machinery spills, hydrocarbon discharge on moving targets.
Satellite imagery analysis to detect Sargassum algae rafts at sea and predict stranding locations and quantities along Caribbean coastlines.