Three commitments

01

Empirical honesty over headline numbers

Every published number is reproducible from on-disk artefacts. Bugs are disclosed, not buried — the May 2026 file_age_days correction is a worked example of this commitment, not an exception to it.

02

Calibration before raw accuracy

The deployed model is the one that ranks honestly, not the one with the highest AUC on a test set. Production selection criteria explicitly privilege calibration, top-k stability, and threshold sensitivity.

03

Governance is not an afterthought

Payload controls, prompt versioning, privacy hashing, and audit trails are part of the architecture, not a compliance bolt-on. Every model call produces a verifiable receipt.

What's next

Three near-term threads that complete the loop from prediction to prevention to attribution.

Next

Telemetry-validated risk lift

Instrument the production deployment so the top-k coverage numbers from Paper C are replicated on real CI telemetry, not just on the offline holdout.

Next

Defect-attribution prototype

Move Paper F from vision to evaluated prototype: fuse stack traces and suspect-set ranking with SHAP-explained priors; user-study practitioners' triage time with vs. without.

Next

Cross-language transfer

Extend Paper B's LORO methodology beyond Java/JS to Python and Go — the languages where the platform's industrial customers actually live.