Marco Franzon

Building practical AI systems in the real world.

About

I build AI systems that run outside the lab on real hardware, in real environments, with real constraints. My focus is on computer vision, edge AI, and physical systems where latency, cost, and reliability actually matter.

I care about making AI practical, reproducible, and understandable. If it only works on a beefy GPU in a notebook, it's not done yet.

Experience

2023 — now

Dualistic

Co-Founder

Digital twins, simulation, and AI for physical systems.

2020 — now

eXact Lab

Data Engineer

HPC infrastructure and machine learning pipelines at scale.

2023 — now

Docker

Docker Captain

Community leadership, containers, and developer tooling.

Projects

Computer Vision with Marco

Open-source org with YOLO training templates, a matrix-vision app, and real-time detection demos.

YOLO Python Computer Vision

Agentic Viz

Visualizer for agentic markdown planner — turn agent plans into interactive diagrams.

JavaScript Agents Visualization

Datacenter Planner

Create your data center from a 2D plan — place racks and cooling systems in a detailed 3D scene.

Three.js 3D JavaScript

Talks & Community

Docker Community All-Hands

Containers for ML workflows — Docker Captain spotlight

Edge AI in Production

Lessons from deploying CV models on resource-constrained devices

Open Source Contributions

Active contributor to Docker ecosystem and ML tooling projects

Writing

Don't buy GPUs, buy microcontrollers 2025
Most AI projects fail in deployment 2025
Make ML boring 2024
Why digital twins need edge inference 2024

Principles

Ship real systems Demos are easy. Production is the work that counts.
Simplicity wins The best solution is the one you can explain in one sentence.
AI runs everywhere If it needs a data center, it's not finished yet.