Architecting Malafaareh — Modern E‑Commerce with AI on GCP Serverless

I designed and built Malafaareh.com, a modern e‑commerce platform focused on performance, reliability, and intelligent customer experiences. This article details the architecture, the AI recommendation/follow‑up system, and how Firebase fits into deployment and user authentication.

Architecture Overview

• Frontend: React/Vite single‑page app served behind a global CDN (Cloud CDN).
• Backend: Serverless API on Cloud Run (Gen2) with autoscaling, containerized Node.js services.
• Data: Cloud SQL (PostgreSQL) for transactional data; Cloud Storage for media assets; Pub/Sub for async workflows.
• Infrastructure: Provisioned via Terraform for consistency and repeatability.
• Observability: Cloud Logging + Monitoring with request tracing and alerting.

AI Recommendations & Follow‑Ups

The platform integrates AI for personalized product recommendations and post‑purchase follow‑ups (email/notifications). A lightweight recommender blends content signals (metadata, categories) with behavioral signals (views, cart events) to produce ranked suggestions. Follow‑ups use event triggers (Pub/Sub) to send context‑aware messages (e.g., complementary items, care instructions, or timely restock alerts).

• Feature embeddings and TF‑IDF for fast retrieval of similar products.
• Contextual rules and business constraints layered over model outputs.
• Throttled messaging with opt‑out compliance and deliverability checks.

Deployment & Authentication (Firebase)

Firebase streamlines deployment and user auth:

• Hosting/Deploy: CI/CD publishes the SPA to Firebase Hosting, which fronts Cloud CDN for low latency.
• Auth: Firebase Authentication handles email/password and social logins with secure session management.
• Integration: Auth tokens gate API routes in Cloud Run; role‑based access controls protect admin tools.

Results

• Sub‑second median page loads on core flows.
• Scalable serverless backend with cost‑aware autoscaling.
• Higher engagement and repeat visits via relevant recommendations.

If you’d like a similar architecture or AI integration, contact me: meandyougtn@gmail.com · LinkedIn: @marubozu.