Model Development on the C3 Agentic AI Platform Overview
The vast majority of production machine learning systems rely on a complex series of data transformation and training operations to serve their final predictions. The C3 Agentic AI Platform allows data scientists to author machine learning workflows of arbitrary complexity through an intuitive authoring framework assembled as Directed Acyclic Graphs (DAGs). The nodes of these DAGs represent units of machine learning work that can be composed of estimators or transformers from leading open-source frameworks, imported models from third-party machine learning service providers (such as, Vertex AI, Azure ML, Sagemaker), or even arbitrary native Python code. This enables data scientists to work directly with native frameworks of their choice, and then simply convert (instead of rewrite) their work to production-ready pipes that are chained together as an MlPipeline.
Benefits of using the tools offered by the C3 Agentic AI Platform for model development:
- Use declarative pipeline authoring
- Share and re-use implementation across projects
- Remove need to write glue code stitching components together
- Track experiments
- Integrate with C3 AI Model Deployment and Model Registry
- Scale and optimize performance:
- Parallelize pipeline operations using C3 Agentic AI Platform's distributed system
- Optimize data transfer operations between nodes of the pipeline DAG during execution
This section of the C3 AI Data Science and Machine Learning Guide includes the following topics and tutorials:
- Model Development Typical Workflow on the C3 Agentic AI Platform
- Machine Learning Pipelines on the C3 Agentic AI Platform
- Author Machine Learning Pipelines - MlPipeline Tutorial
- Author Custom Machine Learning Pipes in Native Python - MlLambdaPipe Tutorial
- Author Custom Machine Learning Pipes in Native Python - MlDynamicPipe Tutorial
- Hyperparameter Optimization at Scale - HPO Tutorial
- Interpretability Deep Dive - ShapInterpreter Authoring and Execution Tutorial