Hepatitis C virus (HCV) infection causes viral hepatitis leading to hepatocellular carcinoma. Despite the clinical use of direct-acting antivirals (DAAs) still there is treatment failure in 5–10% cases. Therefore, it is crucial to develop new antivirals against HCV. In this endeavor, we developed the “Anti-HCV” platform using machine learning and quantitative structure–activity relationship (QSAR) approaches to predict repurposed drugs targeting HCV non-structural (NS) proteins. We retrieved experimentally validated small molecules from the ChEMBL database with bioactivity (IC50/EC50) against HCV NS3 (454), NS3/4A (495), NS5A (494) and NS5B (1671) proteins. These unique compounds were divided into training/testing and independent validation datasets. Relevant molecular descriptors and fingerprints were selected using a recursive feature elimination algorithm. Different machine learning techniques viz. support vector machine, k-nearest neighbour, artificial neural network, and random forest were used to develop the predictive models. We achieved Pearson’s correlation coefficients from 0.80 to 0.92 during 10-fold cross validation and similar performance on independent datasets using the best developed models. The robustness and reliability of developed predictive models were also supported by applicability domain, chemical diversity and decoy datasets analyses. The “Anti-HCV” predictive models were used to identify potential repurposing drugs. Representative candidates were further validated by molecular docking which displayed high binding affinities. Hence, this study identified promising repurposed drugs viz. naftifine, butalbital (NS3), vinorelbine, epicriptine (NS3/4A), pipecuronium, trimethaphan (NS5A), olodaterol and vemurafenib (NS5B) etc. targeting HCV NS proteins. These potential repurposed drugs may prove useful in antiviral drug development against HCV.