Projeto de Machine Learning em formato “production-like” que prediz a probabilidade de sobrevivência de passageiros do Titanic.

Inclui:

• modelo de ML treinado
• API REST com FastAPI
• aplicação containerizada com Docker
• pipeline de CI/CD com GitHub Actions
• Infraestrutura como Código com Terraform

flowchart LR
    subgraph Dev["Developer Workspace"]
        C[FastAPI + ML Code]
        T[Unit/API Tests]
    end

    C -->|git push| R[(GitHub Repo)]

    subgraph CI["GitHub Actions (CI/CD)"]
        A1[Lint & Tests]
        A2[Security Scan]
        A3[Build Docker Image]
        A4[Publish Artifact]
    end

    R --> A1 --> A2 --> A3 --> A4

    subgraph Runtime["Runtime Environment"]
        API[FastAPI /predict]
        Model[(ML Pipeline Artifact)]
        API --> Model
    end

    A4 -->|deploy| API

    subgraph IaC["Terraform (IaC)"]
        TF[Cloud Infrastructure Templates]
    end

    TF -.-> Runtime

• Survival prediction using ML
• REST API endpoint (/predict)
• Input validation
• Docker support
• CI/CD with GitHub Actions
• Terraform templates for cloud infrastructure

curl -X POST http://localhost:8000/predict \
  -H "Content-Type: application/json" \
  -d '{
    "pclass": 3,
    "sex": "male",
    "age": 22,
    "sibsp": 1,
    "parch": 0,
    "fare": 7.25,
    "embarked": "S"
  }'

docker build -t titanic-api .
docker run -p 8000:8000 titanic-api

Open in your browser: http://localhost:8000/docs (Swagger UI)

CI/CD Patterns used:

- container build
- basic validation checks
- packaging for deployment

This ensures every change is validated and reproducible, reducing configuration drift and manual errors.

The infrastructure layer is fully described using Terraform, enabling:

• reproducible environments
• consistent deployments
• cloud-agnostic design principles

This approach reflects production-grade practices commonly used in real-world cloud architectures.

This project demonstrates the candidate’s ability to:

• Design cloud-ready architectures for ML systems
• Bridge Data Science and Backend Engineering
• Implement DevOps automation and Infrastructure as Code
• Think in terms of production systems, not isolated experiments
• Apply software engineering and architectural principles to ML workloads

It reflects practical experience with:

• API design
• containerization
• CI/CD pipelines
• cloud infrastructure modeling
• end-to-end solution delivery

• Functional ML model
• API available for inference
• Dockerized application
• CI/CD pipelines configured
• Terraform templates included
• Study / portfolio project (not a commercial product)