Custom Products & R&D
Product development, industrial demonstrators, R&D in computer vision, embedded AI and material-structure traceability.
From idea to industrial prototype, from proof of concept to scalable product.
Our preferred domains
Computer vision
Defect detection, automated quality control, counting, classification. Custom YOLO model training, from dataset to cloud or edge deployment. See our YOLO model training offering.
Embedded AI
Inference on constrained targets (Jetson, Pi, MCU), quantisation, memory optimisation, accuracy/latency trade-offs.
Traceability & authentication
Unique material signature, anti-counterfeit. Expertise from the QRIDS project (Quick Response Identification by Structure).
Instrumentation & measurement
Multi-sensor acquisition, software test benches, industrial equipment control.
Our method
- Discovery: problem framing, technical constraints, success criteria.
- Pilot (2-4 weeks): feasibility proof with real data, clear go/no-go.
- Agile iterations: 2-week sprints, working deliverable at every sprint end.
- Industrialisation: tests, documentation, knowledge transfer.
Example: QRIDS
Co-founded with James Aubert, QRIDS identifies and authenticates physical parts from their material micro-structure. We designed:
- The QBOX optical device (hardware + firmware).
- The identification algorithms (DFT, PolarSign).
- The full software infrastructure (API, database, containers).
- Demonstrators for industrial partners (watchmaking, luxury).