Advancements in AI, particularly LLMs, are reshaping educational assessment. This session explores how AI can be applied across the full assessment lifecycle to create more efficient, scalable, and responsive systems. We examine AI in item development, where LLMs generate aligned content, stimuli, and rubrics—accelerating timelines and expanding item pools while maintaining human oversight. We then explore AI approaches to item difficulty and discrimination modeling, where fine-tuned LLMs predict item parameters from text and attributes, reducing reliance on field testing and improving calibration efficiency. The session introduces AI-enabled psychometrics, highlighting a shift from static processes to continuous measurement systems. A key theme is human-in-the-loop design to ensure validity, fairness, and transparency. We conclude with an integrated framework connecting item development, scoring, and psychometrics into a cohesive, data-driven ecosystem.