The National Highway-05 (NH-05) stretch from Shimla to Kinnaur in Himachal Pradesh is frequently threatened by landslides, causing disruptions to transportation networks, particularly during severe weather conditions. This vital road, crucial for both local residents and tourists, serves as a lifeline and holds strategic significance, especially with the increasing tourism in the region. In recent years, frequent devastating landslides in the region has resulted in significant loss of lives and extensive damage to vehicles, highlighting the urgent need for a comprehensive landslide susceptibility assessment in the area. This research employs a combination of multi-criteria decision geospatial analysis (MCDGA) and advanced statistical machine learning (ML) models in order to gain a detailed understanding of landslide dynamics in the area. By utilizing the Analytical Hierarchy Process (AHP) and two ML algorithms, Multi-Layer Perceptron (MLP) and Extreme Gradient Boosting (XGBoost), susceptibility models are developed, revealing significant portions of the region categorized as very high and high susceptibility zones. Field surveys supports key factors contributing to increased landslide occurrences, including steep jointed rocks, extensive road cuttings, changes in land use patterns, and various environmental factors. The comparative analysis shows that ML models, including MLP and XGBoost, exhibit superior precision in modelling the landslide susceptibility, with area under curve (AUC) values of 0.948 and 0.955, respectively, compared to AHP( 0.805).Particularly, XGBoost achieves an accuracy of 89.88%, outperforming MLP (87.64%).These findings underscore the effectiveness of ML algorithms in providing precise insights into the susceptibility of the region, enabling more informed and proactive risk mitigation strategies.