Unregulated chromite mining causes enrichment of hexavalent chromium in the groundwater. Due to unpredictable monsoonal recharge and anthropogenic dependencies on groundwater, the depth and extent of chromium pollution becomes extremely difficult to demarcate. For this specific objective, the present study was carried out in order to explore the potential of a coupled surface and sub-surface modelling approach in Sukinda valley, which accounts for 97–98 % of the total chromite reserve of India. Through ionic speciation, saturation state and clustering analysis, the most probable source and corresponding mineral stability state was investigated. In order to trace the extent, status and severity of the problem, both hydrogeologic parameters as well as the geogenic soil parameters were taken into account to develop DRASTIC, DRASTIC-L as well as NOBLES Index. While DRASTIC and DRASTIC-L model provided assessment of vulnerability due to surface leaching of contaminants, NOBLES index, speciation analysis and geochemical model provided sub-surface assessment of vulnerability due to chromium. MRSA and SPSA sensitivity analysis were applied in order to understand the most critical factor that can dominantly control the surface contamination in the groundwater. Random Forest (RF) based machine learning techniques were applied in order to integrate the sub-surface as well as surface characteristics for the purpose of prediction of chromium in the groundwater. The present study therefore presents a novel methodology of risk assessment for regions where either extensive mining activities are operational or in regions with abandoned mines with operative acid mine drainage.