One Weekend: How I Found 3,000 Open-Source AI Projects (and You Can Too)

Description and use

The author mapped 3,000 open‑source AI projects in 48 hours using curated lists, GitHub topic search, dependency graphs, and paper‑to‑code discovery.

Problem

Discovering the breadth of open‑source AI projects efficiently.

Outcome

The author produced a repeatable workflow that yielded ~3,000 open‑source AI projects in 48 hours, revealing that the local‑inference ecosystem is more active than API‑wrapper projects.

Workflow

• Start with curated "awesome-*" GitHub repos to gather an initial list.
• Scrape entries, deduplicate, and compile ~1,200 unique projects.
• Search GitHub topics (e.g., llm, rag, local-llm) to surface maintained projects.
• Use topic:llm stars:>100 to find popular projects and drop star threshold for long tail.
• Explore dependency graphs of trusted tools to find related projects within one or two hops.
• Search GitHub for ArXiv paper titles or referenced repos to capture early implementations.
• Categorize the compiled list into ~30 real categories and identify ~200 actively maintained projects.

Evidence

I gave myself 48 hours to map the open-source AI landscape.

Source

These aren't just lists — each entry is a link with a one-line description. I scraped them into a single file and deduplicated. That alone gave me ~1,200 unique projects.

Source

Searching topic:llm stars:>100 gives you the projects people actually use.

Source

Pick a tool you already trust and look at what it depends on, or what depends on it.

Source

The 3,000 projects collapse into maybe 30 real categories, and maybe 200 that are actively maintained.

Source