Layer provides a central repository for your Datasets and Featuresets so that they can be systematically built, monitored, and evaluated to reduce the cost of errors. Layer Data Catalog with data management, monitoring and search tools, helps data team members to find the data that they need, serves as an repository of available data.
Layer Data Catalog is a collection of metadata which includes many functions to provide a cataloging feature for your whole data. These functions are:
- Data Discovery: With a powerful search, Layer Data Catalog enables easy discovery of your company data and assets while adhering to authorization and governance rules.
- Auto Versioning: With schemas changing, datasets becoming lost, files being deleted and more, pipelines can break, usually taking down models with them. Auto Versioning not only helps you create reproducible ML pipelines but also lets you revert to a previous instance of a model. It helps you pinpoint the data version that leads to the dismal performance.
- Feature Store: Feature engineering is often repeated on new projects as clean source data / features are difficult to discover and access. Layer Data Catalog enables you to declaratively build your features and to serve them in both online (real-time) and offline (batch) fashion.
- Data Quality Testing: Layer makes sure your data meets your quality expectations by executing automated tests.
- Data Lineage: For many organizations, how a feature or model is really created remains a mystery. There are many nested levels of transformations and understanding the full picture is no less than wizardry. Layer Data Catalog does a comprehensive bookkeeping of how different entities depend on each other.
- Close the communication gap on data
- Start new projects quickly
- Solve duplication