We propose a deep learning framework to evaluate CEO-firm compatibility and predict firm performance based on CEO traits and organizational context. Using a feedforward neural network (FNN), we model complex, nonlinear relationships between executive characteristics and future adjusted returns, achieving high predictive accuracy. Our model incorporates interaction terms between CEO attributes and firm-level filters – such as firm value, market-to book ratio, and financial constraints – to capture the contingent nature of CEO impact. SHAP based interpretability reveals that traits like network size, educational prestige, insider status, and management experience drive firm success, but their predictive importance varies across firm environments. The model achieves strong out-of-sample performance with efficient training dynamics and no overfitting, demonstrating the power of machine learning to advance both predictive and theoretical insight in corporate governance research.