Traditional file-based systems store data in separate files managed by individual applications. Although such systems are simple and suitable for small-scale data processing, they become inadequate when data scientists work with large, complex, and continuously changing datasets. Analytics and machine learning workflows require data to be collected from multiple sources, cleaned, integrated, queried repeatedly, and shared among different users and applications. File-based systems create several limitations in performing these tasks efficiently. One major limitation is data redundancy and inconsistency. In a file-based environment, different applications may maintain their own copies of similar data. For example, customer information may be stored separately in sales, marketing, and service files. When the same customer changes an address, updating every copy becomes difficult. If one file is updated while another is not, inconsistent information is produced. Such inconsistencies can negatively affect statistical analysis and machine learning models because algorithms depend on accurate and reliable training data. Another problem is data isolation. Data is often distributed across files created in different formats and locations. A data scientist may need information from customer records, transaction files, website logs,