You signed in with another tab or window. Reload to refresh your session.You signed out in another tab or window. Reload to refresh your session.You switched accounts on another tab or window. Reload to refresh your session.Dismiss alert
Minimum depth bar for Reference Architecture and Case Study submissions used as Due Diligence adopter interview substitutes
Background: The TOC now lets a published TAB Reference Architecture (and soon, a CNCF Case Study) substitute for an individual adopter interview during project Due Diligence. That only works if RA/case study submissions reliably carry interview grade evidence, and today they don't. Asking the TAB to set a minimum depth bar for submissions.
Neither the merged DD guide text nor the TAB's own RA submission process currently requires any minimum depth for a submission to count as adopter interview equivalent evidence, and there's no cap on how many of the required interviews a single project can satisfy this way.
Host:
Note Taker:
Recurring Items
Agenda
Minimum depth bar for Reference Architecture and Case Study submissions used as Due Diligence adopter interview substitutes
Background: The TOC now lets a published TAB Reference Architecture (and soon, a CNCF Case Study) substitute for an individual adopter interview during project Due Diligence. That only works if RA/case study submissions reliably carry interview grade evidence, and today they don't. Asking the TAB to set a minimum depth bar for submissions.
docs(operations): adopter submission form link and TAB reference architecture substitution toc#2256 (merged) added the original TAB Reference Architecture substitution to the TOC's Due Diligence guide: a project's inclusion in a published RA "may serve as a replacement for an individual adopter interview."
docs(operations): allow CNCF Case Study as adopter interview substitute toc#2282 (open) extends the same substitution to CNCF Case Studies, with the same "may serve as a replacement" language.
Neither the merged DD guide text nor the TAB's own RA submission process currently requires any minimum depth for a submission to count as adopter interview equivalent evidence, and there's no cap on how many of the required interviews a single project can satisfy this way.
Should we differentiate or stagger the release of overlapping AI-infra reference architecture submissions ([Reference Architecture] One Heterogeneous Compute Pool, Two Workload Types: How China Merchants Bank Unified Training and Inference on Kubernetes #140 China Merchants Bank vs NB Bank: Building a Full-stack Cloud-native AI Service System Based on Kubernetes and HAMi #141 Bank of Ningbo) before either reaches a publication PR?