Sector tracker · One-engineer companies · Threshold-driven
AI & Machine Learning — the one-person unicorn lens.
Where the one-person research repo turns into a one-person inference company before anyone notices.
Thresholds — AI & Machine Learning
What makes a AI & Machine Learningrepo “solo-founder” here
- Stars (floor)
- ≥ 800
- Commits (rolling 90d)
- ≥ 90
- Distinct contributors
- ≤ 2
- Top-contributor share
- ≥ 92%
Concentration window: rolling 90 days.
Why one founder, why this sector
The shape of one-engineer companies in AI & Machine Learning
AI/ML is the densest solo-founder sector on the panel by an order of magnitude. A single technical founder with a fine-tuned model and a thin inference wrapper can ship a paying product before the first contributor PR lands. The shape we see most often: an evening-and-weekend research repo accumulates stars from a single Hacker News thread, the founder ships an OpenAI-compatible endpoint in week six, and revenue precedes hiring by 8–14 weeks. The hiring lag is the signal — strong commit velocity with no contributor growth means the founder is still alone and still in control of the technical direction.
Tooling footprint
What the codebase looks like
Single-developer signatures: vLLM or llama.cpp in deps, a /api/openai-compatible route, Modal or Replicate in workflow files, a Stripe-checkout-only Next.js front-end, no team-issue templates, README written in first-person singular.
Pattern to watch
The observable acceleration shape
Three consecutive 14-day windows with ≥40% commit-velocity growth while distinct authors stay at 1. Stars climb in step-functions tied to single Twitter/X threads, not gradual organic growth. The repo flips from research-licensed to commercial-licensed mid-quarter.
Most common false positive
What looks like solo-founder signal but isn’t
Cursor-augmented solo developers commit AI-suggested code at velocities indistinguishable from a 3-person team. Verify by inspecting commit message style and the ratio of single-line to multi-line commits — co-pilot output skews short and surgical.
Archetype (composite — not a real person)
The AI & Machine Learning solo-founder shape, in one sentence
Ex-FAANG ML engineer ships an inference proxy over a weekend, gets 4k stars in 72 hours, takes the GitHub Sponsors button down once Stripe is live, and is still the only contributor 90 days later.
Composite archetype. We don’t name founders publicly — that edge belongs to dashboard subscribers, not the open web.
Where the live AI & Machine Learning data lives
From this thesis to the working board
This page is the editorial lens. The live data feeds are next door. /predicted is the weekly all-stage bet. /startups-to-watch ranks this sector by acceleration. /firstlook is the paid Dashboard where the threshold filter actually runs.
Related sectors
Where else this archetype shows up
Related answer: /answers/how-to-find-ai-startups-before-they-raise.
Last reviewed 2026-05-22. Sector entries reviewed monthly. Methodology: /methodology.