Corentin Regal developed and enhanced deployment workflows for two major repositories, focusing on reliability and maintainability. For ggml-org/llama.cpp, he improved the Dockerized workflow by implementing a graceful shutdown mechanism using bash scripting and Docker best practices, ensuring proper signal handling and resource cleanup during container termination. In huggingface/text-generation-inference, Corentin updated the CI/CD pipeline by refining the nix_build.yaml workflow with semantic versioning for Docker images, leveraging YAML and GitHub Actions to improve tag consistency and deployment traceability. His work addressed operational risks, streamlined release processes, and laid the foundation for future automation, demonstrating depth in DevOps engineering.