🌐 English · 🇪🇸 Leer en español

I'm a biologist and genomics specialist (PhD, University of Saskatchewan) based in Canada. I spent 18+ years reading genomes — from Galápagos mockingbirds to black soldier flies — and now I borrow what genomics taught me to build a different kind of AI: one that splits the problem, not the model.

Find your way around

Fragment the problem, not the model.

Most decentralized AI cuts a model across machines, so everything waits for the slowest one. Swarmbly cuts the request: a small local model splits a structured task into micro-tasks, ordinary computers answer them in parallel with small models, and the answers are reassembled on your own machine — the way a genome is assembled from sequencing reads.

flowchart TB
  U(["You"]) --> APP["<b>SwarmblyApp</b><br/>desktop client + network dashboard"]
  APP -->|"your documents stay on this computer"| LCE["<b>SwarmblyLCE</b><br/>local-first personal memory"]
  APP -->|"only if you enable sharing"| NET[("Peers on your network")]
  CORE["<b>SwarmblyAI</b><br/>protocol · spec · research harness"] -.->|"defines how nodes talk"| NET
  classDef app fill:#14405a,stroke:#3cc8dc,color:#fff;
  classDef lce fill:#10303f,stroke:#7bd88f,color:#fff;
  classDef core fill:#10303f,stroke:#e9b949,color:#fff;
  class APP app
  class LCE lce
  class CORE core

📄 Whitepaper: Semantic Fragmentation and Stochastic Assembly (v2, EN/ES, CC BY 4.0) · LCE whitepaper: 10.5281/zenodo.23150478

flowchart LR
  A["✅ Local chat<br/>+ LAN swarms"]:::done --> B["🔧 Stable<br/>grounded answers<br/>privacy tests"]:::next
  B --> C["🧪 Private beta<br/>one-click install"]:::todo
  C --> D["📘 Public beta<br/>protocol v1.0-rc<br/>conformance tests"]:::todo
  D --> E["🌐 Internet tier<br/>rendezvous · NAT"]:::todo
  classDef done fill:#1d5a3a,stroke:#7bd88f,color:#fff;
  classDef next fill:#14405a,stroke:#3cc8dc,color:#fff;
  classDef todo fill:#2a2f36,stroke:#8b949e,color:#fff;

A quick tour of what is new in my work, from the lab bench to the swarm.

Smaller projects from the lab and the field, all public on my GitHub.

Eight peer-reviewed articles in population genomics, molecular ecology and conservation — and a PhD on Galápagos mockingbirds.

xychart-beta
    title "Peer-reviewed articles per year"
    x-axis [2020, 2021, 2022, 2023, 2024, 2025, 2026]
    y-axis "Articles" 0 --> 3
    bar [1, 2, 0, 1, 2, 1, 1]

🎓 PhD thesis (2023): Population Genomics of the Galapagos Island Mockingbirds and Implications for Conservation — University of Saskatchewan.

Preprints & open-source software

🔗 All publications: Google Scholar · ORCID · ResearchGate

gantt
    title Career timeline
    dateFormat YYYY-MM
    axisFormat %Y
    section Education
    BSc Biology - PUCE           :2004-09, 2010-12
    MSc Conservation Biology - PUCE :2012-01, 2016-12
    PhD Biology - USask          :2017-09, 2023-05
    section Teaching and research
    Lecturer - PUCE              :active, 2014-08, 2026-10
    PhD researcher and TA - USask :2017-09, 2023-05
    section Industry
    NRGene Canada - NAIC         :2022-11, 2026-09
    section Independent
    Swarmbly                     :active, 2026-08, 2026-10

Swarmbly is independent, open source (AGPL-3.0) and not funded by any institution. Support pays for the first real-world tests — physical computers and a rendezvous server — and for the time to publish the results, including the negative ones. It doesn't buy tokens, shares or promised features.

_{Thanks for stopping by. If anything here is unclear, write to me — explaining things well is part of the work. 🧬}