SYSTEM Cited by 1 source
Netflix AIP Portal¶
The AIP Portal (AI Platform Portal) is Netflix's unified UI surface for exploring the MDS model lifecycle graph. It provides full-text search across all entity types, detailed entity pages with navigable relationships, and personalized views for teams and individuals.
First documented in the 2026-05-04 Netflix TechBlog post sources/2026-05-04-netflix-democratizing-machine-learning-building-the-model-lifecycle-graph.
Typical interaction shape¶
"A typical interaction in the AIP Portal looks like:
- Search: Type a model, feature, dataset, or team name into the single search box backed by Elasticsearch.
- Inspect: Land on an entity page that shows key metadata (description, owners, domains, tags) alongside a relationships panel.
- Explore: Click through to related entities (upstream datasets, downstream experiments, and sibling model versions) to navigate the Model Lifecycle Graph without leaving the portal."
The two-store split underneath (patterns/dual-store-graph-plus-search-index) maps directly to the UI flow:
| User action | Backing store |
|---|---|
| Type a name in the search box | systems/elasticsearch |
| Click into an entity page | systems/netflix-mds |
| Click a related-entity link | systems/datomic (graph traversal via MDS) |
Investigative workflow example (from the post)¶
"Imagine an engineer investigating a degraded engagement metric for a personalization model. They might:
- Start with the model instance powering the affected recommendations in the AIP Portal.
- Inspect the model's features and follow a suspicious feature to its upstream dataset.
- From the dataset page, see that its pipeline recently had failed runs and identify the owning team.
- Confirm which A/B tests are currently running this model instance to understand which members and surfaces are impacted."
Pre-MDS this was a 4-system manual walk. Post-MDS it is a contiguous click-through journey in a single UI.
Auto-generated entity-type baseline¶
"When new entity types are introduced into MDS, the portal automatically provides baseline search, entity pages, and relationship navigation, and we can then layer on domain-specific visualizations (such as model deployment history or dataset version timelines) over time."
The portal treats search + entity pages + relationships panel as a generic baseline UX shape that any entity type gets for free, with domain-specific UIs (deployment history, lineage timelines) layered on top per-entity-type.
Staleness disclosure¶
The portal surfaces the last-enriched timestamp for each entity:
"newly discovered relationships may appear with a short delay after the underlying entities are created (typically minutes rather than seconds). We track when each entity was last enriched and surface this timestamp in the AIP Portal, so practitioners can reason about staleness and know when it's safe to rely on a particular relationship for debugging or impact analysis."
This honesty about staleness is rare in catalog UIs — the lag between underlying state change and graph visibility is exposed as first-class data instead of hidden inside the pipeline.
Open UI challenges (per the post)¶
- Domain-specific visualizations: "Different entity types require distinct visualization experiences. Model pages should display deployment history, A/B test associations, and performance metrics. Feature pages should highlight data lineage and consuming models. Pipeline pages must show execution history, dependencies, and schedules. Dataset pages require versioning timelines and downstream consumers. How do we design a flexible UI framework that allows each entity type to have its own tailored experience while maintaining consistent navigation and interaction patterns across the portal?"
- Risk if unaddressed: "the portal risks becoming a generic catalog rather than a tool that ML practitioners rely on in their daily workflows."
What is not documented¶
- Rendering framework / web stack.
- Deployment topology / traffic shape.
- User count / per-page latency.
- Whether the portal exposes a public-API surface in addition to the UI.
- Access-control / per-team views beyond the "personalized views for teams and individuals" mention.
Seen in¶
- sources/2026-05-04-netflix-democratizing-machine-learning-building-the-model-lifecycle-graph — canonical introduction.