Loop engineering is a skill now being absorbed into AI tools, distinct from prompt engineering, which concerns extracting one good response from an AI model. Modern coding assistants such as Claude Code, Cursor and GitHub Copilot have moved beyond single-shot interaction, with developers configuring loops that repeatedly call an AI agent on a schedule to work through a backlog unattended, repair broken builds, respond to review comments and report progress. Loop Engineering is the discipline of designing such workflows responsibly, addressing what triggers a run, what the agent may access, how much it may spend, and when it must pause and consult a human. The field gained relevance because running an agent in a loop has become inexpensive and dependable enough to be trusted with genuine work. The discipline is expanding into Graph Engineering as organisations move from a single autonomous agent to several collaborating agents. No dedicated degree exists, but foundations include a B.Tech or B.E. in Computer Science, Information Technology, or AI and Data Science, postgraduate study in AI and machine learning systems, an MCA with an AI focus, or an M.Tech centred on agents or systems, plus short certifications in DevOps, CI/CD, cloud infrastructure and applied agentic AI offered this year by Anthropic, Google and Indian ed-tech providers. Entry-level work increasingly involves supervising loops that write and review code, and roles such as AI agent orchestration engineer, autonomous systems engineer and applied AI engineer already exist at product companies and consultancies building AI-augmented engineering teams.