APEX: Automated Prompt Engineering eXpert with Dynamic Data Selection

APEX (Automatic Prompt Engineering eXpert) is a newly introduced framework designed to enhance prompt optimization for Large Language Models (LLMs) by dynamically selecting and stratifying datasets into Easy, Hard, and Mixed tiers based on performance. It significantly improves data efficiency, achieving an average performance increase of 11.2% on the Gemini 2.5 Flash model and 6.8% on the Gemma 3 27B model across benchmarks like IFBench and SimpleQA Verified, emphasizing the importance of a data-centric approach in prompt engineering. This advancement is crucial for practitioners aiming to optimize LLM performance while minimizing computational resources.