Edvise

Edvise is a school-agnostic Python library for implementing Student Success Tool workflows. It provides shared components for preparing institutional data, training and running predictive models, and producing reports for student-success interventions.

Edvise is designed primarily for workflows orchestrated on Databricks. Using the complete pipelines requires institution-specific configuration, data, and cloud infrastructure.

What Edvise provides

• Data ingestion from Google Cloud Storage into Databricks and Unity Catalog
• Schema validation and data-quality auditing
• Feature generation, student selection, and outcome target creation
• H2O model training, calibration, evaluation, registration, and inference
• Model monitoring and drift detection
• MLflow-backed model cards and reporting
• GenAI-assisted identity and schema mapping with human review
• Databricks Asset Bundles for deploying data, modeling, and reporting jobs

Requirements

• Python 3.10, 3.11, or 3.12
• uv for dependency management

Running production pipelines may also require:

• A Databricks workspace with Unity Catalog
• Databricks CLI authentication and appropriate workspace permissions
• Google Cloud credentials and access to the required GCS buckets

Local setup

Clone the repository and install the locked development environment:

git clone https://github.com/datakind/edvise.git
cd edvise
uv sync --frozen --dev

Run the test suite:

uv run python -m pytest

Run the local quality checks:

uv run ruff check .
uv run ruff format --check .
uv run python -m mypy src

Some tests and runtime paths depend on external services and may require additional credentials or configuration.

Using Edvise

Edvise is organized around configurable workflows rather than a single command-line interface. A typical implementation:

1. Maps institution data to the expected schema.
2. Selects and adapts a configuration template from configs/.
3. Configures a Databricks Asset Bundle from pipelines/.
4. Deploys and runs the relevant ingestion, training, inference, and reporting jobs in Databricks.

Useful starting points include:

• PDP H2O configuration templates
• Legacy H2O configuration template
• GenAI mapping configuration
• H2O training entry point
• H2O inference entry point

Configuration and infrastructure vary by institution. Treat the templates as starting points rather than complete deployment configurations.

Repository structure

.
├── configs/       # Workflow and institution configuration templates
├── notebooks/     # Legacy workflow notebook templates
├── pipelines/     # Databricks Asset Bundles
├── src/edvise/    # Python package and job entry points
└── tests/         # Automated test suite

The package is divided into modules for ingestion, data auditing, feature generation, modeling, reporting, student selection, synthetic data, and GenAI mapping.

Contributing

Before opening a pull request:

1. Add or update tests for behavioral changes.
2. Run the test suite, Ruff checks, and mypy.
3. Update CHANGELOG.md when the change is relevant to a release.

Use GitHub issues to report bugs or propose features.

License

Edvise is available under the Apache License 2.0.