The mission of BOSS AI is to enable every leader in every organization to put their data to work no matter where it resides. The BOSS Modeling Framework (BMF) enables developers to build and interface custom AI models with the BOSS Unity Client platform for streamlined management, experimentation, and training. Data scientists use the Unity client interface for managing data and parameters associated with models and experiments.
The framework supports the development of python-based AI models built with TensorFlow, PyTorch, Scikit-learn, and XGBoost (with the Dask parallel computing module). BMF's python libraries support the following tasks:
- Accessing BOSS virtual datasets (VDS) for model training and evaluation
- Analyzing and reporting model performance metrics, such as confusion matrices and ROC curves
- Storing structures representing trained models and training checkpoints
BMF also supports distributed model training using the Horovod framework (https://horovod.ai/).
The BOSS platform simplifies the integration process with the following features:
- Real-time data ingestion from existing systems
- Scale extract, transform, and load (ETL) workflows to any data size
- Interconnect structured, unstructured, and semi-structured data quickly
- Integrations for connecting with most major enterprise systems, enabling users to rapidly integrate the BOSS platform into existing IT infrastructures and applications.
- Transform data for any AI/ML workloads
- Unify data across silos in a multicloud environment
- Enforce data-level security models for all data and meet data governance requirements "out of the box."