Building tools for constraint-first AI architecture — starting with L-BOM 💘.
We focus on privacy-preserving, locally-run tools that can be used to analyze and optimize the compute requirements of large language models. We are in early development, but we are excited to share our work and collaborate with others in the space.
| Repo | Description | Status |
|---|---|---|
| L-BOM 💘 | LLM Bill of Materials Generator | v0.2.0 |
| GUI-BOM | GUI Implmentation of L-BOM | v0.2.0 |
| HissCheck 🐍 | AI Powered Python Testing Validation | v0.1.0 |
| Repo | Description | Status |
|---|---|---|
| Ridge Sight ⛰️ | Cross-Repo Pull Request Management Dashboard for GitHub | ✅ Production |
- Privacy first — local-network, no cloud dependency
- Constraint-first design — compressed compute, ternary weights
- Open research — transparent architecture decisions
We allow the use of AI code generation tools (e.g. GitHub Copilot) to assist in development, but we require that all code generated by such tools be fully reviewed by a human before being committed to the repository. This is to ensure that all code meets our quality standards and does not introduce any unintended issues.
Submission of code generated by AI tools should be marked in the pull request description, and the reviewer should verify that the generated code is appropriate and does not contain any security vulnerabilities or other issues.
In addition, pull requests should be limited in scope to a single concern, to make it easier for reviewers to understand the context and implications of the changes being made. This helps maintain the quality and integrity of our codebase while still allowing for the benefits of AI-assisted development. Pull requests that are too broad or that contain multiple unrelated changes may be requested to be split into smaller, more focused pull requests.
We strongly encourage those learning how to code using the help of AI tools to utilize the fork function of GitHub to create a personal copy of the repository where they can experiment and learn without the risk of affecting the main codebase. This allows for a safe and supportive environment for learning and growth while still contributing to the community. Pull requests from forks are welcome and will be reviewed with the same standards as any other contribution.
