Design and Automation of a Modular ETL Workflow Using Workato: A Practical Approach to Scalable and Reusable Data Pipelines

In the context of modern data engineering, the ability to build scalable, maintainable, and low-code ETL (Extract, Transform, Load) workflows is becoming increasingly critical. This thesis presents the design and implementation of a modular ETL framework using Workato, a cloud-based integration and automation platform (iPaaS). The proposed workflow orchestrates data extraction from a PostgreSQL source hosted on Supabase, applies structured transformations and loads the results into multiple target tables, while enforcing business rules and referential integrity. The solution is structured into two pipeline levels (L0 and L1), which reflect the progressive stages of data preparation and validation, and leverages Workato Recipes and Recipe Functions to ensure modularity and reusability. An external Python script has been developed to automatically adapt the .JSON recipes to new data schemas, enabling the pipeline to be easily reused across different datasets. Design choices such as schema separation (l0, l1, admin), variable parameterization, and error management are discussed to highlight scalability and maintainability. Although the integration of AI functionalities is suggested for future work, this project lays a solid foundation for low-code ETL automation and demonstrates the flexibility of Workato in supporting real-world data pipelines.