whoami
Yannick Roy, PhD
Head of AI @ Almanak · AI × Blockchain x Quant Leadership
I lead multidisciplinary teams that turn ambitious AI research into products people trust—across DeFi, quantitative finance and beyond. I designed the agentic framework behind $200M+ AUM in strategies. I manage AI-Native teams of Humans and Agents.
TARGET ROLES AI Leadership · Product Management · Engineering Leadership — across AI, Blockchain & Quant
psst — try for the terminal version `
// CAPABILITIES
Multi-agent orchestration that survives contact with real capital — not demos. Multi-model routing across leading providers, evals, guardrails, and the ML pipelines underneath (ETL → train → serve).
Smart contracts to strategy execution: Solidity, web3.py, Anvil, Tenderly — shipped across 15+ chains and the protocols that matter.
PhD in neuroscience (closed-loop BCI systems). Author of the first Deep Learning for EEG review — 2,000+ citations. I treat strategies like experiments: hypothesis, test, reproduce.
I ship code weekly and run the roadmap. Founded and scaled NeuroTechX to 19,000+ members in 30+ chapters across the world. Teams I build are fast, ego-less, and user-obsessed.
// AI-NATIVE.REPS
A working bench I use heavily to compare models, route work, and keep humans and agents moving.
I evaluate leading models on real product work, then use the best fit for research, engineering, and review.
// AI.JOURNEY
Before AI became the defining technology of our time, it was already the defining thread of my work.
I develop my first multilayer perceptron—my first hands-on step into neural networks.
The most interesting class I had ever taken, inside one of the communities shaping modern deep learning.
I publish the first systematic review of deep learning for EEG, with Alex Gramfort and collaborators.
2,000+ CITATIONS
We republish the updated MOABB: the Mother of All BCI Benchmarks, an open-source framework for reproducible BCI research. Launched in 2018, it's still having a great impact in the field!
I build a multi-agent team—strategist, quant, coder, QA, and reviewer—before agentic workflows become the default playbook.
12 AGENTS IN PRODUCTION
KitchenLoop turns intent into tested software—and helps build an advanced open-source SDK for DeFi.
OPEN SOURCE
// PATTERN.MATCH
The pattern is not simply time spent in AI. It is repeatedly working on consequential ideas before they become consensus.
ML AND DL
AND BCIs
AND DEFI
AND AI TEAMS
RESULT SEEING THE SIGNAL BEFORE THE WAVE.
// HISTORY.LOG
DeFi stack for quants. Own everything AI from PoC to production — shipped the agentic framework powering $200M+ AUM, coordinating R&D, infra, engineering, data, and UX.
Built the largest international neurotech community: 19,000+ members, 30+ chapters, 300+ events.
Built a neurotechnology consulting and recruiting company. Exited in 2022.
Closed-loop cognitive training systems. Deep learning on EEG — noisy, non-stationary time-series. The experimental discipline transfers: hypothesis, test, reproduce.
Real-time and embedded systems in C/C++ — flight simulators, telecom, industrial 3D vision across 10+ countries.
// FIELD.NOTES
A decade of building communities, teaching, and speaking — 300+ events organized through NeuroTechX, lectures in advanced C programming at ÉTS, and panels from research venues to policy roundtables.
The same loop every time: learn in the weeds, build with the field, teach what works. That loop is now pointed at AI agents for markets.
// PROJECTS.LS
// RESEARCH.BIB
2,000+ CITATIONS · FIRST DL-FOR-EEG REVIEW
AI-NATIVE SOFTWARE · SELF-EVOLVING CODEBASES
PHD THESIS · UNIVERSITÉ DE MONTRÉAL · 2015–2022
Full list on Google Scholar · ResearchGate — or run in CLI mode
// CONTACT.INIT
// DIGITAL.TWIN.POC
A live 3D prototype grounded in Yannick's approved site profile. It is an AI—not the real Yannick.
Camera is optional. Audio and video stay in your browser; only the question text is sent to /api/ask.