EngineersofAI LogoCoursesBlogResearch LabAI LettersThe LabCode BankInteractive 3DKodrEarnestJobs
Skip to main content
Interactive 3D/Model Registry & Version Lifecycle
Model Registry - fraud-detector
Version tree with stage lifecycle. Click a version to inspect lineage. Promote models through stages.
Version
Trained
Accuracy
F1
Latency
Stage
Promote To
v1
2025-11-01
0.871-5.0%
0.854
48ms
Archived
v2
2025-12-15
0.904-1.7%
0.891
52ms
Archived
v3
2026-01-20
0.921
0.914
45ms
Production
v4
2026-03-10
0.934+1.3%
0.928
61ms
Staging
Lineage - fraud-detector v3
Training Data
data-v2.0.0
Code Commit
i7j8k9l
Base Model
xgboost-v2
Stage Legend
ArchivedRetired versions
CandidateUnder evaluation
StagingPre-prod testing
ProductionLive in production
Display
Key Insight
The registry is the single source of truth for which model is live. Each version links to its exact training data, code commit, and parent model - enabling reproducibility and rollback.

Model Registry & Version Lifecycle - Interactive Visualization

A model registry is the central repository for managing trained model versions. Each version moves through a defined lifecycle: Candidate (newly trained), Staging (pre-production testing), Production (serving live traffic), and Archived (retired). The registry stores full lineage - which training data, code commit, and base model produced each version. When a model degrades in production, you can instantly identify the exact data version to re-train from.

  • Click Promote buttons to move a model version through Candidate → Staging → Production
  • Production stage automatically archives the previous production version
  • View full lineage for any version: training data version, code commit hash, base model
  • Accuracy delta shows improvement over current production for each candidate version

Part of the EngineersOfAI Interactive 3D - free interactive visualizations covering every major concept in machine learning and AI engineering. Hover any element for a plain-English explanation. No code required.