Use the product
The public app is live: source search, Linked Art records, IIIF imagery, rights context, and evidence links are reachable without a mock layer.
Product case study
Meta Museum turns cultural heritage APIs into a usable public web app: researchers can explore records, inspect provenance and rights, review AI-assisted outputs, and trace every launch claim back to an evidence artifact.
Evaluator brief
Watch the 90-second demo. No-narration recording; the replacement script lives in the docs.
What matters in 90 seconds
Meta Museum is best judged as a live standards product with unusually explicit evidence boundaries: the useful app is already there, and the remaining 10/10 work is named instead of glossed over.
The public app is live: source search, Linked Art records, IIIF imagery, rights context, and evidence links are reachable without a mock layer.
The system is a Next.js SaaS surface with provider adapters, org-aware storage, AI review gates, ActivityStreams, and executable readiness artifacts.
The portfolio story is strong now; strict 10/10 production proof still waits on real Update/Delete feed reads, 30-day SLO/uptime, and one invoice-backed pilot.
Public deployment evidence, launch review, uptime probes, k6 SLO runs, docs freshness, and review-goals audits are generated artifacts rather than hand-written claims.
Three external ActivityStreams consumers and three verified callbacks are recorded. The remaining honest gap is type coverage for real reviewed Update rows and real upstream tombstone Delete rows.
The managed pilot path has buyer packs, manual invoice entitlement support, activation evidence, support tracking, KPI packets, and guardrails. It still needs a real signed pilot before it counts as SaaS traction.
14
Registry-backed source lanes with preserved provenance and rights context.
1286
App, API, protocol, docs, privacy, launch, and evidence checks run in the local gate.
8/10
Public demo is strong; strict production 10/10 waits on external proof.
Live
Demo, architecture story, and evidence links are visible without weakening buyer credibility.
What I built
This is not only a visual demo. It is a full-stack SaaS product surface for cultural heritage data: standards modeling, public UX, AI assistance, operational proof, and commercial pilot scaffolding in one deployable Next.js application.
Provider adapters preserve source payloads while normalizing records into Linked Art JSON-LD, HAL links, ARK identifiers, IIIF image review, and graph/entity pivots.
Rights, citation freshness, originality, sensitivity review, AI refusals, and human approval gates keep public claims and generated text grounded in source evidence.
The app ships with deployment preflight, DR rehearsal, k6 SLO packets, public uptime probes, GA4 consent posture, review-goals audits, and runbooks for pilots and external feed adoption.
Live surfaces
Each route is backed by the current provider registry, app services, tests, and launch evidence rather than static project placeholders.
A registry-backed provider map for collection APIs, search/import capabilities, credential requirements, and provenance-aware source review.
Review sourcesExplore imported records as source-backed Linked Art JSON-LD, with graph pivots, entity pages, pattern discovery, and IIIF image review.
Explore recordsPrepare reviewed publication drafts with citation templates, rights warnings, source-of-truth checks, and human approval gates.
Read bridge designArchitecture proof
The public story stays short, but the proof is still reachable: source capabilities, review controls, runbooks, and strict-gate artifacts remain linked for deeper technical review.
Plain-English status for product readiness, deploy readiness, SaaS proof, and remaining external evidence.
Open roadmapStructured launch, pilot, long-term SLO, ActivityStreams, consent, and Render-service evidence gates.
Open auditSingle-page architecture, discipline, state, and evidence summary for the full engineering story.
Open summaryHonest readiness
The app is credible as a professional web project because the live product, demo, architecture, tests, deployment evidence, and limitations are all visible. The production 10/10 gate remains intentionally stricter: it still requires 30 clean days of SLO/uptime evidence, one real invoice-backed pilot, and real ActivityStreams Update/Delete read coverage from external consumers.