Arnaud Bletterer
Research Scientist & Team Leader
Computer Graphics · Image/Geometry Processing · Machine Learning
Bridging the gap between exploration and industrialization.
Professional Profile
Selected Employment
Quartier Latin
Leading core technical research, algorithm design, and high-performance pipelines for advanced computer vision systems.
QuantifiCare & Quartier Latin
Led and mentored a research team of 7. Professionalized workflows, automated validation pipelines, and supervised clinical research publications.
QuantifiCare
Validated and industrialized robust and autonomous multi-view 3D reconstruction algorithms for daily clinical software products.
Core Skills
Education
Université Côte d'Azur (2018)
Visual Demonstrations
Partial Voronoi Diagrams
An interactive guide to Voronoi partitioning on partially known domains. Visualize Voronoi diagrams, Lloyd's relaxation (CVT) and local safety envelope basins for local Voronoi diagrams computations.
Open demo
The Bilateral Filter
An interactive guide to edge-preserving smoothing. Visualize how spatial closeness and color similarity combine to filter noise while keeping boundaries sharp.
Open demo
Point Cloud Normal Estimation
An interactive guide to normal estimation on point clouds. Visualize neighborhood graph computation, covariance PCA ellipse mapping, and orientation propagation via Minimum Spanning Trees.
Open demoArticles
Shrink the Loop, Keep the Gate
Why teams must compress their planning and delivery cadence to match agent speed, while keeping verification gates fixed so that faster loops do not simply ship mistakes sooner.
Failing Fast, Failing Small: The Courage to Narrow Your Scope
Why attempting to cover every conceivable edge case makes failure catastrophic, and how embracing flow code with limited scope lets teams fail small and correct course effortlessly on up-to-date reality.
The Hubris of Expertise: Why We Must Check the Hypothesis First
Why experienced professionals routinely skip basic reality checks to pursue complex, unneeded work: exploring the universal trap of expert overconfidence across industry, politics, and engineering.
The Architecture of Calibrated Autonomy: Contracts, Supervisors, and Deterministic Oracles
Why true system ownership means mastering architectural intent rather than syntax micromanagement: combining upfront contracts, supervisor-worker hierarchies, and deterministic oracles.
The Mentorship Paradox: Raising Senior Engineers in the Age of Autonomous Agents
Why eliminating mechanical entry-level tasks creates an existential training void, and how R&D leaders must replace routine coding with failure injection, Socratic reviews, and architect-apprentice-agent pods.
The Cadence of Thought: Reclaiming the Time to Reflect
How delegating mechanical execution to AI assistants and autonomous agents liberates the cognitive space for deep engineering reflection, turning public writing into a natural byproduct of unhurried thought.