Devin Pellegrino
AI Systems Architect · Founder, Nerority
Mechanical engineer by training, systems architect by practice. I spend my days deploying FDA-cleared medical technology and building AI agent systems that hold up in production — and I publish the methods as I go.
How I Got Here
Every phase built the next — hardware discipline, then regulated operations, then AI systems.
Engineering Foundations
B.S. in Mechanical Engineering from Georgia Tech — first-principles thinking, systems discipline, and the habit of building things that have to actually work.
Medical Technology & Public AI Work
Joined Brain Ultimate as Director of Engineering — FDA-cleared TMS systems, national deployments, clinician training. Same year, started publishing AI prompt-engineering work publicly; the community formed around it.
Formalization
The public work became a discipline: a formal curriculum, 33 GPT agents, 16 prompt libraries, 13 frameworks — all released and dated. 560+ GitHub stars of external validation.
Production Architecture
Designed the complete prompt and agent architecture for Pet Portal AI — later selected into the Plug and Play Topeka accelerator with Hill's Pet Nutrition. Founded Nerority LLC.
The NERO System
The method matured into a live operating system: 37 persistent domain agents across six planes under apex orchestration, with the full methodology published in the NI-NOTION series.
What I Do Day-to-Day
Director of Engineering
Technical operations for an FDA-cleared TMS medical-device company: installations, nationwide clinician training on deployment and plasticity mechanics, and the internal knowledge systems that keep a regulated operation coherent.
Architectural Engine Lead
The complete prompt and agent architecture behind the pet-health advisory product — grounding, structure, and consistent behavior for a consumer AI with zero margin for confident wrong answers.
Founder & Lead Architect
Consulting, the NERO Method, and a community of 500+ builders — everything I learn running production systems gets published, taught, and pressure-tested by people using it on their own domains.
The NERO Method
Most AI systems fail in open-ended domains because they're designed top-down: generate an architecture, force behavior into it, watch it drift. I work in the opposite direction. Master the domain manually first. Reverse-engineer the implicit expertise into explicit structure. Let the AI attempt operations at full complexity, fail, and get corrected at the exact point of divergence — each correction becomes a permanent rule. The result is agent systems that compound instead of drift, in domains where the only ground truth is human mastery.
See the method running live
Get in Touch
No booking widgets, no forms. If you want to work together, message me directly — email or Discord.