DevelopmentAutomation and AI

Docker infrastructure for a customer service AI agent, delivered early and ready for the server

A developer building an AI agent to answer customers on Instagram needed the project's foundation in place: the services starting together, with health checks and the same environment on a laptop and on the server.

Client: André M.
Who did it: Lucas Barreto

Problem

The client's plan called for three services (a Node gateway, a Python AI engine and a vector database) and asked for a Node version that was already out of support. Without a base that starts with one command, every later phase of the agent would begin by fighting the environment.

Solution

A two-layer Docker Compose setup, one safe for the server and one for development, with the 3 services, a health check on each, non-root users and pinned dependencies. The Node version was moved to the current LTS, explained in the pull request, and acceptance was checked again from a clean copy of the client's repository.

Result

All 3 services start with one command and pass their health checks. The pull request was opened on day one and delivery came 3 days before the agreed deadline.