Water sources near mining regions are often susceptible to contamination from toxic elements. This study employs machine learning (ML) techniques to evaluate drinking water quality and identify pollution sources near a chromite mine in Iran. Human health risks were assessed using both deterministic and probabilistic approaches. Findings revealed that concentrations of calcium (Ca), chromium (Cr), lithium (Li), magnesium (Mg), and sodium (Na) in the water samples exceeded international safety standards. The Unweighted Root Mean Square water quality index (RMS-WQI) and Weighted Quadratic Mean (WQM-WQI) categorized all water samples as 'Fair', with average scores of 67.95 and 67.19, respectively. Of the ML models tested, the Extra Trees (ET) algorithm emerged as the top predictor of WQI, with Mg and strontium (Sr) as key variables influencing the scores. Principal component analysis (PCA) identified three distinct clusters of water quality parameters, highlighting influences from both local geology and anthropogenic activities. The highest average hazard quotient (HQ) for Cr was 1.71 for children, 1.27 for adolescents, and 1.05 for adults. Monte Carlo simulation for health risk assessment indicated median hazard index (HI) of 4.48 for children, 3.58 for teenagers, and 2.98 for adults, all exceeding the acceptable threshold of 1. Total carcinogenic risk (TCR) exceeded the EPA's acceptable level for 99.38 % of children, 98.24 % of teenagers, and 100 % of adults, with arsenic (As) and Cr identified as the main contributors. The study highlights the need for urgent mitigation measures, recommending a 99 % reduction in concentrations of key contaminants to lower both carcinogenic and non-carcinogenic risks to acceptable levels.