Logo

Monitoring

Model Health Command Center

Production Health

Active Models

193

Critical Models

2

Business Impact Score

94

AI Generated Summary

193 production models are healthy overall. Churn Prediction v3.2 and Sentiment Analysis v1.4 are flagged critical due to accuracy drift and feature drift respectively. Retraining is recommended within 48 hours for both.

Model Health Matrix

Health score by model · last 7 days

247 models tracked · 193 in production View registry

Prediction Accuracy Trend

Churn Prediction v3.2 · last 5 weeks

Fleet-wide forecast accuracy 97.8%

Feature & Data Drift

Distribution shift by feature

Reviewed by Priya Patel · ML Platform

Inference Latency Analysis

p50 / p95 · last 24 hours

18.4M predictions served today 99.98% uptime

Resource Utilization

GPU / CPU / Memory / Storage

Owned by Marcus Lee · Infra

AI Explainability

Churn Prediction v3.2's accuracy decline correlates most strongly with a shift in the "weekly_active_sessions" feature distribution following a product UI change 3 weeks ago.

weekly_active_sessions0.42 shift
avg_session_duration0.19 shift
Analysis by Dana Ruiz · Data Science

Model Incident Timeline

Churn Prediction v3.2 flagged critical

2 hours ago

Sentiment Analysis v1.4 flagged critical

5 hours ago

Fraud Detection v4.1 passed weekly audit

1 day ago

2 critical incidents open

Scheduled Retraining

Churn Prediction v3.3

Scheduled tonight, 1:00 AM

Sentiment Analysis v1.5

Scheduled tonight, 2:30 AM

Owned by Jordan Kim · ML Ops

Health Alerts

Churn Prediction v3.2 · accuracy driftCritical
Sentiment Analysis v1.4 · feature driftCritical
Recommendation Engine v2.7 · latency spikeWarning
3 open alerts · 99.98% API availability

AI Recommendations

Retrain Churn Prediction v3.2 tonight using the last 90 days of data to correct drift before it impacts churn interventions.
Sentiment Analysis v1.4 feature drift correlates with new product review categories — refresh the training vocabulary.
2 recommendations pending review

Model Audit History

Fraud Detection v4.1 passed weekly audit1 day ago
Revenue Forecasting v2.3 passed weekly audit2 days ago
Recommendation Engine v2.7 passed weekly audit3 days ago
193 active models audited weekly