3 projects
scicode-lint
AI-powered linter for scientific ML code. Detects methodology bugs such as data leakage, incorrect cross-validation, numerical instability, and reproducibility issues that produce plausible but wrong results. 66 detection patterns across 5 categories. Runs via local or institutional vLLM server, no cloud APIs.
sciwrite-lint
A linter for scientific manuscripts — reference verification, consistency checking, and structural validation.
research-mentor
AI-powered research mentor that guides you through your research journey.