Problem
Customer-success and RevOps teams need actionable churn signals, not just raw model probabilities.
Live Product
ChurnLab
Live multi-tenant SaaS for churn intelligence, explainable ML, and retention operations
PythonFastAPINext.jsTypeScriptPostgreSQLRedisDockerLinuxscikit-learnSHAPPlaywright
Solution
Founded, architected, and launched a production SaaS that ingests customer data, trains churn models asynchronously, scores account risk, explains predictions, surfaces revenue exposure, and supports retention workflows.
What I Built
Multi-tenant authentication and authorization, asynchronous Redis-backed scoring, PostgreSQL persistence, explainable ML, risk dashboards, dataset ingestion, retention actions, Stripe integrations, observability, automated backups, CI/testing, and production operations.
Production Engineering
Operate the live application with Docker, Linux, Cloudflare Tunnel/TLS, health checks, structured logging, automated backups, queue recovery, rate limiting, security controls, and end-to-end testing.