How jo120424/a-ai-ess-guardian works
AI-ESS GUARDIAN is a research prototype for predictive electronic-component burn-in screening. Reliability engineers submit telemetry through the screening dashboard; validated data is assessed for lot-relative anomalies, projected degradation and safety-envelope breaches, then fused into a PASS/REVIEW/REJECT disposition with QA explanations. Separate research workflows compare screening methods on NASA capacitor data and produce validation analyses. The README documents the screening modules and research capabilities; wiring for unsampled implementation details is shown only where documented.
Main entry points: dashboard.py QA Dashboard · processed Processed Dataset · run_experiment.py Comparative Runner · run_validation.py Validation Runner
Components 25
Component Screening
- QA Dashboard
app/dashboard.py - Lot Anomaly Detection
backend/screening/module_a.py - Drift Forecasting
backend/screening/module_b.py - Safety Envelope
backend/screening/safety_envelope.py - Decision Fusion
backend/screening/decision_engine.py - QA Explanations
backend/screening/explainer.py - Training Orchestrator
backend/screening/train_and_persist.py - Artifact Manager
backend/screening/artifact_manager.py - Model Artifacts
models
Telemetry Data
- Telemetry Validation
backend/data_pipeline/sih_pipeline.py - Synthetic Burn-in Data
data/synthetic - NASA Stress Data
data/raw - NASA Preprocessing
backend/data_pipeline/preprocess.py - Processed Dataset
data/processed - Dataset Profiling
backend/data_pipeline/profile_dataset.py
Research Experiments
- Comparative Runner
backend/experiments/run_experiment.py - Ground Truth Labels
backend/experiments/ground_truth.py - Experiment Results
results/step4
Validation & Audit
- Validation Runner
backend/validation/run_validation.py - Validation Analyses
backend/validation - Validation Results
results/step5
Research Web Interface
- Research Frontend
frontend/js/app.js - HTTP API Server
backend/server.py
Other components
- Reliability Engineer
- Research User
Connections 21
- Reliability Engineer → QA Dashboard: submits telemetry
- QA Dashboard → Telemetry Validation: validates input
- Telemetry Validation → Lot Anomaly Detection: clean telemetry
- QA Dashboard → Drift Forecasting: requests forecast
- QA Dashboard → Safety Envelope: checks safety
- QA Dashboard → Decision Fusion: fuses assessments
- QA Dashboard → QA Explanations: requests reasons
- Training Orchestrator → Artifact Manager: persists models
- Artifact Manager → Model Artifacts: reads/writes
- Synthetic Burn-in Data → QA Dashboard: supplies demo data
- NASA Stress Data → NASA Preprocessing: provides source data
- NASA Preprocessing → Processed Dataset: writes dataset
- NASA Stress Data → Dataset Profiling: profiles dataset
- Processed Dataset → Comparative Runner: provides records
- Comparative Runner → Ground Truth Labels: generates labels
- Comparative Runner → Experiment Results: writes results
- Processed Dataset → Validation Runner: provides data
- Validation Runner → Validation Analyses: runs analyses
- Validation Runner → Validation Results: writes reports
- Research User → Research Frontend: opens interface
- Research Frontend → HTTP API Server: fetches API data