Grounded AI and enterprise QA research across retrieval, verification, citation, and abstention.

• Foundation: product, architecture, governance, and research direction
• Eval: datasets, labeling rules, metrics, and benchmark reports
• Retrieval: parsing, chunking, hybrid retrieval, and reranking
• Verification: claim splitting, support checks, contradiction checks, and abstention
• App: API, prompts, structured responses, and user experience
• Ops: CI/CD, dashboards, and release and operations runbooks

• Most implementation repos are private by default.
• Cross-repository planning is tracked in private Projects for delivery and evaluation gates.
• Ownership is split between @ai-infosec-lab/core-maintainers, @ai-infosec-lab/research, and @ai-infosec-lab/platform.

• Evidence before fluency
• Evaluation before release
• Structured citations and uncertainty handling
• Reproducible experiments and clear release gates

• Use pull requests against main for changes that should be reviewed.
• Keep experiments reproducible and record evaluation impact in the relevant repo.
• See the organization-level CONTRIBUTING.md, SECURITY.md, and SUPPORT.md files for shared guidance.