The LLM Engineer's Schedule (kwojcicki.github.io)

A new perspective on the LLM engineer's schedule emphasizes the collaboration between individual contributors, or "makers," and managers within the realm of AI/ML development. The traditional conflict between these roles, where focused, uninterrupted work is vital for makers while managers operate through quick task-switching, is alleviated with the advent of Large Language Models (LLMs). These LLMs enable developers to automate mundane tasks and allow them to engage in meetings without losing productive time. Consequently, the emphasis shifts from a rigid 9-5 work schedule to a more flexible approach, where engineers can run autonomous agents for extended periods, potentially enhancing overall productivity. The significance of this shift lies in the expectation for engineers to adapt their skills and methodologies to harness LLMs effectively. Engineers must now define success criteria and develop a robust process for initiating and monitoring agents, focusing on high-value tasks to avoid catastrophic mistakes from lack of context or poor prompts. As LLMs increasingly handle repetitive work, engineers are encouraged to dedicate their prime hours to deep work and critical decision-making, while automating lower-cognitive-load tasks. This evolving workflow not only reshapes daily routines but also highlights the increasing need for financial foresight as AI costs rise, compelling engineers to balance between utilizing powerful models and managing budgets efficiently.