This paper focuses on the application of data analysis technology based on intelligent computer systems in financial technology product pricing. It addresses the issues of traditional pricing, such as reliance on experience, high-dimensional and nonlinear features, and risk assessment bias caused by sample distribution drift and cold starts. By developing a pricing modeling framework that integrates machine learning and causal inference, the paper addresses these issues. The steps include (1) data governance and feature engineering, including missing multiple imputation, time window aggregation, and domain binning; (2) sample drift detection (population stability index) and reweighting correction; (3) model integration, using gradient boosting decision tree (GBDT) and extreme gradient boosting (XGBoost) for nonlinear fitting, and superimposing elastic net to ensure interpretability; (4) causal effect estimation (double machine learning) to identify the marginal impact of price on conversion and risk; (5) reinforcement learning (RL) to conduct personalized optimal price search and set supervision and fairness constraints; (6) online/offline A/B joint evaluation and model monitoring. Taking a consumer credit product as an example, when the FPR increases to 0.5, the TPR increases to 0.92; the mean absolute error decreases to 0.073; the average pricing yield increases to 0.115; the average high-risk default rate decreases to 0.180.