How rudikcloud/ai-incident-intelligence works
This is an incident-triage ML lifecycle: controlled failure scenarios drive telemetry collection and labeled dataset generation; validated measurements become engineered features for baseline and trained root-cause/incident-type predictions. Evaluation compares both approaches, while the FastAPI endpoint returns top-k rankings, confidence, and feature evidence. The graph shows the documented workflow and source-backed calls; deployment infrastructure and several collector internals were not sampled, so no additional runtime links are inferred.
Main entry points: evaluate.py Model evaluation · generate_dataset.py Scenario collector · app.py Inference API · build_features.py Feature engineering
Components 14
Telemetry collection
- Scenario collector
src/collect/generate_dataset.py
Dataset and features
- Incident dataset
data/incidents.csv - Schema validation
src/features/dataset.py - Feature engineering
src/features/build_features.py
Prediction models
- Heuristic baseline
src/models/baseline.py - Model training
src/models/train.py - Model bundle
Evaluation
- Model evaluation
src/eval/evaluate.py - Evaluation metrics
src/eval/metrics.py - Evaluation report
reports/eval.md
Inference API
- Inference API
src/serve/app.py - Triage results
Other components
- Failure platform
- API client
Connections 18
- Failure platform → Scenario collector: supplies telemetry
README.md - Scenario collector → Failure platform: injects failures, runs workload
src/collect/generate_dataset.py - Scenario collector → Incident dataset: appends labeled rows
README.md - Scenario collector → Schema validation: uses schema columns
src/collect/generate_dataset.py - Incident dataset → Schema validation: validated by
src/features/dataset.py - Schema validation → Feature engineering: provides features
src/features/build_features.py - Feature engineering → Model training: prepares training features
src/models/train.py - Model training → Model bundle: saves models
src/models/train.py - API client → Inference API: submits incident
README.md - Inference API → Feature engineering: builds request features
src/serve/app.py - Inference API → Model bundle: loads models
src/serve/app.py - Inference API → Triage results: returns rankings and evidence
src/serve/app.py - Incident dataset → Model evaluation: evaluated against
src/eval/evaluate.py - Model evaluation → Feature engineering: loads engineered data
src/eval/evaluate.py - Model evaluation → Heuristic baseline: scores heuristic predictions
src/eval/evaluate.py - Model evaluation → Model training: loads or trains models
src/eval/evaluate.py - Model evaluation → Evaluation metrics: computes metrics
src/eval/evaluate.py - Model evaluation → Evaluation report: writes results
README.md