Abstract: To address issues in the experimental teaching of the Data Science and Big Data Technology major—such as the lack of engineering background, virtualized data sources, and “black-box” algorithmic models—this paper presents the design and development of an intelligent flight trajectory prediction experimental platform. Guided by the Outcome-Based Education (OBE) philosophy and leveraging resources from research projects and invention patents, the platform adopts a B/S architecture. It integrates core functions including real ADS-B message parsing, time-series data governance, heterogeneous deep learning model integration, and interactive visual monitoring. In terms of instructional design, a three-stage progressive experimental system consisting of “Basic Verification – Comprehensive Application – Advanced Innovation” is constructed. The “Environmental Attention Encoding” technology derived from scientific research is transformed into feature engineering operators for teaching, guiding students through a full-process engineering practice ranging from data cleaning and feature extraction to Transformer-based model construction and visual decision support. Application practice demonstrates that this platform effectively breaks down barriers between courses such as Data Acquisition and Preprocessing, Data Mining, Time Series Analysis, Data Visualization , and Data Visualization Practice. It achieves a deep transformation of research resources into teaching resources, significantly enhancing students’ systematic thinking and innovative capabilities in solving complex engineering problems.