Layer is a Declarative MLOps (DM) platform to help data teams across all companies to produce machine learning applications based on code.
With DM, you define what to accomplish, rather than describe how to accomplish it. You pass your Dataset, Feature and ML model definitions, and Layer builds your entities seamlessly. This enables you to focus on what matters to you - designing, developing and deploying models - without worrying about the infrastructure to do it.
- Abstraction: Datasets, Featuresets and ML Models are first-class entities in Layer. You can easily monitor and manage their lifecycle which will not just streamline but also simplify your MLOps.
- Versioning: Layer tightly couples and versions your Data, ML Model and Code. It provides extensive UI functionality to do version-diffing not on the underlying code but also with your Data and ML model versions.
- Reusability: We provide two central repositories which contain reusable entities across all projects within your organization:
- Data Catalog: The data catalog is a place you can discover and access trusted datasets and features so that you can generate impactful insights that drive business outcomes. All datasets and features are immutable and versioned.
- Model Catalog: The purpose of the model catalog is to provide a managed and centralized storage space for ML Models. It ensures that model artifacts are versioned and immutable. It allows data teams to manage and monitor the lifecycle of the ML Models.
- Scalability: Layer not only introduces infra agnostic pipelines to maximize your resources for high scalability but also helps you manage the lifecycle of your Data and ML models at scale.
- Reproducibility: Layer automatically tracks lineage between your versioned entities. Even if your entities change over time (datasets, features) you will be able to reproduce the experiment with the same results which will provide transparency and give you confidence in understanding exactly what was done.
- Observability: Layer not only introduces traceable processes which provide full observability but also helps you monitor post-production lifecycle of your entities. For example, you can track the training metrics of your ML Model along with the model's impact on your business KPIs.
- Continuous Delivery & Deployment: You can create automated or reactive pipelines with Declarative MLOps to automatically build, test, and deploy your Data and ML Models at scale
We built Layer to empower data teams. It can be used by:
- Data scientists to develop machine learning models and track experiments
- ML/Data Engineers to streamline machine learning operations
- Data/Business Analysts to develop a single source of truth and high quality features