The integration of artificial intelligence technologies into pharmaceutical research is showing promising results in both effectiveness and efficiency. Developing an early understanding of absorption, distribution, metabolism, excretion, and toxicity (ADMET) endpoints is crucial to the success of drug de-sign projects. In this study, we present an innovative automatic machine learning platform, named ArtemisAI. The proposed platform offers a fully automated end-to-end pipeline, encompassing feature preprocessing, training and validation, model tuning, and eventual model selection, all conducted with-out human intervention. We employ the proposed platform for in silico ADMET endpoint prediction, and benchmark it against four state-of-the-art baseline models/platforms on the Therapeutic Data Commons (TDC) ADMET datasets. In 17 out of the 22 examined ADMET datasets, ArtemisAI stands out as a top-performing approach, demonstrating superior performance over these instances. For the remaining instances, ArtemisAI consistently ranks within the top two performers. The experimental outcomes showcase the effectiveness and robustness of ArtemisAI on diverse ADMET endpoint prediction tasks.