Incident Ingestion Form
Paste incoming IT ticket payload to execute instant multi-task triage.
Quick Presets:
Loading presets...
AI Triage Engine Ready
Select a quick preset above or type an IT incident to view real-time classification, SLA breach probability, and resolver routing.
High-Throughput Batch Ticket Triage
Upload CSV datasets (up to 1,000 incident tickets) for automated bulk classification.
Drag & Drop Incident Tickets CSV
File must contain title and description columns (Max 1,000 rows)
Total Triaged
--
Real-time
High SLA Breach Risk
--%
Target resolution alerts
Model Confidence
--
Mean multi-class score
Mean Latency
-- ms
Per ticket inference
Category Classification Breakdown
Priority Distribution
7-Day Ticket Triage Velocity
Prediction Confidence Spread
Operator Review Queue (< 0.60 Confidence)
Tickets automatically flagged for second-line human validation.
| ID | Incident Summary | Category | Priority | Confidence | Risk Band | Ingested |
|---|
Production Model Registry & Quality Gate
Active MLflow production alias models deployed for inference.
| Task | Registered Model | Version | Algorithm | Accuracy | Macro F1 | Status |
|---|
Dataset Fingerprint & Provenance
Dataset SHA256: 50b8a12483f92b90
Holdout Test Gate: Category Macro F1 ≥ 0.80 | Priority Macro F1 ≥ 0.60 | SLA ROC-AUC ≥ 0.70
End-to-End MLOps Pipeline
1. Data Ingestion
8k Tickets (EDA + Stratified 70/15/15)
2. MLflow Tracking
12 Runs (Params, Metrics, Artifacts)
3. Quality Gate
Holdout Test Set Validation
4. Model Registry
Aliases: Staging → Production
5. HF Deployment
Dockerized FastAPI + SQLite
Candidate Architecture Benchmark (MLflow Run Comparison):
