AutoFlip Intelligence — portfolio summary
Production-style MVP for used car dealer intelligence: multi-tenant web app with inventory, costs, margins, and ML-based price recommendations.
Delivered application: A production-style MVP for used car dealer intelligence—a multi-tenant web app where dealers manage inventory, track costs, analyze margins, and get ML-based price recommendations.
Tech stack: Frontend — Next.js 14 (App Router), TypeScript, Tailwind CSS, shadcn/ui, Recharts, Lucide React. Backend — FastAPI (Python 3.11+), SQLAlchemy, Pydantic, JWT + bcrypt. Database — PostgreSQL 15, Redis 7 (caching). ML — LightGBM, quantile regression, SHAP explainability, conformal prediction. Infra — Docker Compose, Alembic migrations.
Notable portfolio points: Full-stack Next.js frontend and FastAPI backend; ML in production with price recommendations, confidence intervals, and feature importance (SHAP); multi-tenancy with organization-based isolation and RBAC; data modeling for vehicles, expenses, listings, sales, KPIs, and analytics; production-oriented design with caching, migrations, API docs, and structure suitable for deployment.
