Corentin Regal developed and enhanced deployment workflows for two major open-source repositories over a two-month period. For ggml-org/llama.cpp, he improved Docker container reliability by implementing a graceful shutdown mechanism using bash scripting and Docker best practices, ensuring proper signal handling and resource cleanup during runtime. In the huggingface/text-generation-inference repository, Corentin updated the CI/CD pipeline with semantic versioning for Docker images, leveraging YAML and GitHub Actions to enforce consistent image tagging and improve deployment traceability. His work focused on maintainability and operational safety, addressing core DevOps challenges and laying the foundation for automated version verification in production environments.