⚡ Enterprise CLV & Churn Inference Engine

🟢 Live API Dashboard: Test the live Swagger UI here

🚀 Overview

An enterprise-grade Machine Learning Operations (MLOps) REST API serving dual serialized XGBoost models. This system evaluates multi-dimensional customer metadata to predict churn probability and Customer Lifetime Value (CLV), coupled with automated heuristic business triage recommendations.

✨ Enterprise MLOps Features

• Native C++ TreeSHAP Explainability: Sub-millisecond feature attribution extracting the specific positive/negative drivers (e.g., support ticket volume) behind every real-time prediction.
• Vectorized Batch Inference: Dedicated /predict/batch endpoint engineered for high-throughput, multi-record array processing.
• Zero-Latency Async Telemetry: Non-blocking FastAPI background tasks that silently stream all API traffic and inference latency metrics to a local SQLite audit ledger.
• OOD Anomaly Guardrails: Strict out-of-distribution payload validation and boundary checking via Pydantic v2.
• Automated CI/CD Test Suite: End-to-end pytest coverage verifying latency thresholds, explainability schemas, and boundary rejections.

🧠 System Architecture

1. Data Ingestion: Enforces type checking on input payloads.
2. Dual-Model Inference:
   • Churn Classification: Predicts risk percentage based on support cadence and CSAT.
   • CLV Regression: Forecasts lifetime revenue value.
3. Actionable Triage Engine: Maps risk scores to operational actions (e.g., Critical Risk intervention vs. Enterprise Upsell).

💻 Local Execution

1. Install Dependencies

pip install -r requirements.txt