MLOps Repository

This repository contains materials and code for MLOps and LLMOps case studies, including infrastructure setup, API development, model training, and prompt engineering workflows.

Main Folders

• aula2_mlop_infra/

  • MLOps infrastructure, API, and model training workflows
  • api/: Flask API for model serving
  • docker/: Dockerfiles, Compose, and requirements
  • notebooks/: Jupyter notebooks for ML experiments
  • src/: Model training scripts

• aula3_case_study/

  • LLMOps case study and workflow orchestration
  • airflow/: DAGs, configs, and logs for Airflow pipelines
  • api/: API and LLM application code
  • docker/: Docker setup for Airflow and API
  • notebooks/: Jupyter notebooks for prompt engineering and MLflow
  • src/: Prompt engineering, model validation, and training scripts
  • tests/: Unit tests for API and infrastructure

Automation & CI/CD

• GitHub Actions: Automated checks for clean Jupyter notebooks, code quality, and reproducibility
  • Ensures notebooks are free of outputs and execution counts before merging
• Dependabot: Automated dependency updates for Python and Docker
  • Keeps dependencies secure and up-to-date

Getting Started

1. Clone the repository:

   git clone https://github.com/luismaiaDEVSCOPE/mlops.git

2. Review the README files in each folder for specific setup instructions.
3. Use Docker Compose files for environment setup as needed.

Requirements

• Python 3.8+
• Docker & Docker Compose

Useful Links

• MLflow UI: http://localhost:5001
• JupyterLab: http://localhost:8888
• API: http://localhost:8080

License

MIT License.