TransmogrifAI is an AutoML library for building modular, reusable, strongly typed machine learning workflows on Apache Spark with minimal hand-tuning.
The library addresses the challenge of reducing machine learning development time while maintaining model quality. It automates feature engineering, algorithm selection, and hyperparameter tuning through a domain-specific language built on Spark. The approach achieves accuracies close to hand-tuned models while reducing development time by approximately 100x. The API enforces compile-time type-safety, modularity, and reusability, allowing developers to build production-ready machine learning applications in hours rather than months.
TransmogrifAI suits teams working with structured data on Spark who want to accelerate model development without requiring deep machine learning expertise. It is particularly valuable for organizations seeking to build machine learning applications quickly while maintaining code quality and reusability. The library provides both a high-level automated interface for rapid prototyping and lower-level control for developers who want to specify features and algorithms explicitly. It integrates as a standard dependency into existing Scala and Spark projects.
The project maintains active development with regular updates to support current versions of Spark and Scala. The codebase demonstrates sustained engineering effort, with documentation and examples kept current alongside the core library. The project includes comprehensive scaladoc API documentation and a dedicated documentation site with getting started guides, examples, and frequently asked questions.
