epic: Spring 2026 AASCU convening follow-ups

[William-Hill]

Summary

Tracks all engineering follow-ups from the AASCU Intermediary feedback session (Spring 2026 convening). Source: docs/aascu_intermediary_feedback_summary.md.

Goals (from the convening)

1. Close the data trust gap — make every number provably traceable to source rows.
2. Meet the AI/governance bar — FERPA-plus expectations on transparency, lineage, and sensitive-population safeguards.
3. Reduce institutional process fragility — submission knowledge that lives with one person, opaque PDP feedback, broken handoffs.

Child issues

#	Title	Size
#105	Metric definitions glossary with IPEDS / state cross-walks	M (3–5d)
#106	Presentation-ready chart export (PNG/PDF) with definitions baked in	M (4–6d)
#107	Data lineage view — "where did this number come from"	L (8–14d)
#108	AI transparency page — model inventory, data flow, provider disclosure	S (1–2d)
#109	Sensitive-population safeguards — feature exclusion + low-N warnings	M–L (5–8d)
#110	Upload validation report — row-level errors, coercions, diff vs. last upload	M (4–6d)
#111	Submission runbook generator — capture and replay institutional steps	M (4–6d)
#112	Spike: institution grouping helper for fall datathon	XS (1d)

Rollup: ~30–48 engineering days.

Recommended sequencing

1. feat: AI transparency page — model inventory, data flow, and provider disclosure #108 first — cheapest, highest procurement-leverage; ship as a forcing function that surfaces gaps in feat: data lineage view — "where did this number come from" #107 and feat: sensitive-population safeguards — feature exclusion lists and context warnings #109.
2. feat: metric definitions glossary with IPEDS and state-compliance cross-walks #105 — content-heavy; unblocks feat: presentation-ready chart export (PNG/PDF) with definitions baked in #106.
3. feat: data lineage view — "where did this number come from" #107 + feat: upload validation report — row-level errors, coercions, and diff vs. last upload #110 + feat: submission runbook generator — capture and reproduce institutional upload steps #111 as a bundle — they all need the same upload-event / provenance schema. Building once and writing all three against it is materially cheaper than three separate efforts. The sandbox epic (epic: SIS → PDP → AR → Dashboard sandbox #116) is the natural substrate for proving these out.
4. feat: presentation-ready chart export (PNG/PDF) with definitions baked in #106 — depends on feat: metric definitions glossary with IPEDS and state-compliance cross-walks #105 for definition strings.
5. feat: sensitive-population safeguards — feature exclusion lists and context warnings #109 — touches ML pipeline + per-institution config; lands after the trust/transparency layer is in place.
6. spike: institution grouping helper for fall datathon (shared SIS + shared goal) #112 — research spike; can run in parallel with anything.

Related

• epic: SIS → PDP → AR → Dashboard sandbox #116 — SIS → PDP → AR → Dashboard sandbox (substrate for feat: data lineage view — "where did this number come from" #107 / feat: upload validation report — row-level errors, coercions, and diff vs. last upload #110 / feat: submission runbook generator — capture and reproduce institutional upload steps #111)
• docs/aascu_intermediary_feedback_summary.md — source feedback

[William-Hill]

Started epic via #105 (metric glossary): branch feat/issue-105-metric-glossary adds /glossary, content/metric-glossary.md, KPI tooltip deep links, and coverage tests. Next per sequencing: expand glossary to charts/other pages, then #107/#110/#111 bundle on sandbox (#116).

[William-Hill]

Update: Chart views now have glossary anchors and full entries (#risk-alert-distribution, #retention-risk-funnel, #readiness-assessment). Branch feat/issue-124-glossary-chart-links — open PR when ready. Completes the #135 follow-up for in-context definitions on dashboard charts.