Thesis: Automating Pipeline and Cloud Deployment Using Large Language and Large Action Models (LLM-LAM)
This research aims to merge LLMs' language understanding with LAMs' action execution capabilities, creating a robust tool for DevOps and cloud operations. The solution will support automated pipeline creation, cloud deployment, and infrastructure management, contributing to enhanced efficiency and scalability in deployment workflows.
Description
This thesis explores how Large Language Models (LLMs) and Large Action Models (LAMs) can be integrated to automate complex workflows, focusing on the creation of CI/CD pipelines and cloud infrastructure deployment. By leveraging these advanced models, the system will be able to manage and execute tasks across platforms such as Jenkins, GitLab, and various cloud environments. Automation methods will include generating code and interacting with APIs, streamlining the deployment process for diverse applications.
Key Features
-
Pipeline Automation
- Utilize LLMs to generate pipeline configurations, such as YAML scripts for GitLab CI/CD.
- Translate user inputs into executable code that automates tasks including environment setup, model training, and deployment.
-
Cloud Deployment via APIs
- Deploy LAMs for handling large-scale actions such as infrastructure provisioning and cloud deployment using tools like Pulumi.
- Enable automatic resource management and deployment across different cloud platforms.
Challenges
- Ensuring accuracy in translating natural language instructions into precise technical commands for automation.
- Integrating diverse platforms, APIs, and deployment tools seamlessly into the automation workflow.