Nathan Leroy

TL;DR - Computational Biologist, ML Engineer, and Builder

  • 5th-year PhD in working with transformers for single-cell genomics.
  • Experienced ML engineer, Rust + Python developer, and product builder.
  • Open-source author: Atacformer, Textractor, PEPhub.
  • Co-founded novacards.ai: helping 1,000+ med students study smarter.
  • Looking for ML/AI research, bioinformatics engineering, or founding engineer roles.
  • Available Winter 2025 — and open for consulting engagements now!

What I Work On

🧬 Atacformer – Foundation model for scATAC-seq

Transformer-based model for single-cell ATAC-seq. Enables zero-shot cell-type annotation & clustering directly from raw fragment files.

  • 10x parameter efficiency vs existing tools
  • Written in PyTorch with custom Rust tokenizers and backends
  • Preprint: bioRxiv 2023.08.01.551452
  • 👉 GitHub Repo | Demo Notebook | Docs

🧠 NovaCards.ai – AI-powered study tool for med students

SaaS platform using generative AI and semantic search to create personalized Anki flashcards.

  • React/Next.js frontend, Supabase backend, OpenAI LLMs, Stripe billing
  • Built Anki plugin with JWT-based login system
  • 1,000+ users, thousands of cards indexed
  • 👉 Live site | Blog post