Peer to Peer (P2P) lending allows individuals and institutions to bypass traditional financial intermediaries and loan money to borrowers through online platforms. This fosters accessibility and efficiency, but inherently carries risks like credit default and financial loss for lenders. The burgeoning Indian P2P lending industry is projected to reach $10.5 billion by 2026, driven by technological advancements and government support. Our project delves into this dynamic landscape, leveraging machine-learning to refine credit risk assessment methodologies. Traditionally, boosting algorithms were often considered “black boxes” due to their complexity, hindering interpretability. We address this challenge by integrating sophisticated model interpretability techniques like SHAP (Shapley-additive-explanations) and LIME (local-interpretable-model-agnostic-explanations), etc. Our approach utilizes regression models to predict loan default probability and classification algorithms to categorize borrowers into different risk categories. We will orchestrate ensemble learning techniques, integrating algorithms such as LightGBM, Random Forest, and ExplainableBoosting(ebm), among others, to construct a resilient and adaptable framework.