I build artificial intelligence systems from the ground up. Not chatbots. Not wrappers. Not integrations. Complete architectures—neural networks, training pipelines, inference engines, autonomous agents—designed for organizations that need absolute control over their AI infrastructure. My work spans nuclear fusion control, industrial automation, medical imaging, and enterprise systems that cannot afford to depend on external APIs or surrender their intellectual property to cloud providers.

Algoretico — Building Systems That Matter

Since founding Algoretico in 2018, I've developed AI architectures that solve real problems in demanding environments. Our innovations include:

Talents — A breakthrough concept introducing persistent neural layers that shape how AI models learn, specialize, and excel in specific domains

Maestro — The first European patented architecture for orchestrating autonomous AI agents

You Need No Data — A framework for self-supervised AI training that allows training AI models with minimal or no data at all

SPaaS — Software Production as a Service, transforming how organizations build software through AI-driven automation

These aren't concepts. They're deployed systems: controlling proton-boron fusion reactors at millisecond precision, diagnosing pathologies in medical scans, orchestrating industrial workflows in steel manufacturing, and powering enterprise RAG systems with proprietary knowledge that never leaves the organization's infrastructure.

Research at the Frontier

My research explores the boundaries between theoretical possibility and engineering reality. I've published work on autonomous control for nuclear fusion, GDPR-compliant concept erasure in large language models, adaptive meta-domain transfer learning, and novel training paradigms that challenge conventional assumptions about what neural networks require to learn.

Projects like PoAItry (the first Italian poetry neural network), DNART.io (AI art from human genome sequences), and systems for emotion recognition and industrial sensor fusion demonstrate that serious AI research doesn't require choosing between technical depth and creative exploration.

Teaching the Next Generation

I teach Machine Learning, Deep Learning, and Natural Language Processing at three universities across Italy:

• Università Vita-Salute San Raffaele (Milan)

• Istituto Italiano di Criminologia (Vibo Valentia)

• Università Europea di Roma (Rome)

My approach combines mathematical rigor with hands-on system design. Students learn to architect neural networks, not just import them. They understand backpropagation by implementing it, not by watching videos. The goal isn't to create AI users—it's to create AI builders.

Advisory Work

I advise organizations at the intersection of AI, energy, and industrial transformation: the European Fusion Association on AI integration in fusion research, Lagiste23 as CTO evaluating high-impact technology investments, and Institut EuropIA on European AI ethics and innovation initiatives. The work involves translating frontier AI capabilities into practical systems that operate in regulated, high-stakes environments.

Books

I've written three books exploring AI from different perspectives—technical foundations, societal implications, and creative applications:

• Alchimia (2024) — On AI, creativity, and transformation

• Intelligenza Artificiale e Criminologia (2023) — AI in criminology and justice systems

• La Matematica dell'Intelligenza Artificiale (2023) — Mathematical foundations of AI

Contact

Email: ciao@michelelaurelli.it

Website: www.michelelaurelli.it