Inference Foundry is an open-source organization building tools and research around LLMs, diffusion models, and efficient local inference. Public pages provide high-level context and project discovery; member-level planning and detailed ownership stay in private org docs.
Organization site: inference-foundry.github.io
Public project index: docs/projects
Detailed member handbook and roster: .github-private (org members only)
| Project | What it is | Links |
|---|---|---|
| super-ollama | Terminal-native, in-process local LLM engine (no HTTP in the main UX); llama.cpp via CGo, focus on low overhead and clean teardown. |
Repo · Roadmap (wiki) |
| Crucible | Open research journal and experimental log. | Repo |
| BitForge (planned) | Quantization theory, methods, and reproducible experiments across bit-widths and runtimes. | Repo TBD — org doc |
| Lexicon (planned) | Open fine-tuned prompt catalog with versioning, licensing, and analysis for reuse. | Repo TBD — org doc |
| Argus (planned) | Algorithms to detect AI-generated images using JEPA-based representations. | Repo TBD — org doc |
For deeper project ownership, plans, and internal notes, use .github-private (members only).
- Kritarth Dandapat — GitHub · LinkedIn · Discord:
kritarth2006 - Atshal Ahmed Khan — GitHub · LinkedIn · Discord:
atshal123 - Community Discord: discord.gg/R8cgA4RpDV
If you want to collaborate or reach the team, contact us via Discord or founder links above.
- Contributing (org-wide): CONTRIBUTING.md
- Code of conduct: CODE_OF_CONDUCT.md
- Security reporting: SECURITY.md
- Community: Discord — architecture, papers, and tooling discussion.
Inference Foundry — open tools and honest measurements.