Welcome to Jukebox, a collection of scenario notebooks and pipelines designed to guide you through the AI/ML lifecycle—from data exploration to production-ready deployments.
The content is organized into the following directories:
- 1-data_exploration: Explore and preprocess the dataset, and set up the feature store.
- 2-dev_datascience: Develop a simple model to test inference on the dataset.
- 3-prod_datascience: Build production-ready pipelines for continuous training, validation, and inference.
- 4-metrics: Set up TrustyAI to monitor data drift and model bias.
- 5-data-versioning: Use DVC to version the dataset for reproducible training.
- 6-advanced_deployments: Experiment with autoscaling and advanced deployment strategies.
- 7-feature_store: Configure Feast (Feature Store) and use it for real-time inference.
- 8-securing_ai: Explore security tools for AI/ML workloads.
- 99-data_prep: Prepares raw data before the lab. This is not used during the lab but can be utilized by the instructor to refresh the dataset beforehand.