Alejandro Saavedra Ruiz

Design, development, and operation of an applied AI backend system deployed in a real production environment, which has evolved through multiple technical iterations from an initial functional implementation into a more robust and maintainable architecture.

The project started as a real application aimed at validating actual usage in production, which required decision-making under real constraints such as cost, complexity, time, and reliability. Throughout this process, I assumed full responsibility for the system: backend design, runtime behavior, environment configuration, and operational stability.

Across successive iterations, I integrated applied AI capabilities as part of the backend, treating AI as an engineering component rather than an isolated feature, and designed task-oriented pipelines with explicit attention to reliability, latency, and cost control. Hands-on production experience made the limitations of the initial architecture clearly visible.

Based on this diagnosis, I decided to deliberately refactor the system, strengthening separation of responsibilities, domain boundaries, and long-term maintainability, using the project as personal engineering work, not as an active business initiative.

Main technical stack

Backend: Node.js (JavaScript / TypeScript), Express, Hexagonal Architecture (Ports & Adapters)

Database: Firebase / Firestore

Applied AI: OpenAI API, Gemini API

Payments: Stripe (checkout and webhooks)

Deployment: Render

Version control: Git, GitHub

Architecture extension (in progress): Python microservice (FastAPI) for RAG and AI orchestration

Secondary stack (production)

App / Frontend: Flutter (Dart) — minimal interface focused on backend consumption and validation

Distribution: Google Play Store