v0.1.0

Geronimo v0.1.0 Release Notes

The first public release of Geronimo — an ML deployment framework that takes you from local development to production with one command.

Core Framework Components (Developer API)

Data Layer

• DataSource — Unified interface for Snowflake, Postgres, SQL Server, and file-based data loading
• Query — Parameterized SQL queries with file-based templates

Feature Engineering

• Feature — Declarative feature descriptors with type hints, transformers, and encoders
• FeatureSet — Fit/transform interface for feature pipelines with automatic column mapping
• derived_feature_fn — Custom functions for multi-column derived features

Model Management

• Model — Abstract base class for wrapping ML models with train/predict lifecycle
• HyperParams — Grid search-enabled hyperparameter configuration
• ArtifactStore — Versioned artifact storage with local and S3 backends
• MLflow integration — Optional backend for experiment tracking

Serving

• Endpoint — Abstract class for FastAPI endpoints with preprocess/postprocess hooks
• AuthMiddleware — API key and JWT authentication middleware
• APIKeyManager — Create, validate, and revoke API keys with scope-based access control

Batch Processing

• BatchPipeline — Abstract class for scheduled batch jobs
• Schedule — Cron and interval-based scheduling (daily, weekly, custom)
• Trigger — Event-based triggers (S3 upload, SNS message)

Monitoring

• ModelSnapshot — Capture feature distributions and model metrics
• Drift detection — Statistical tests for feature and prediction drift

Configuration

• GeronimoConfig — Pydantic-validated project configuration
• Validation rules — Resource sizing, naming conventions, scaling limits

---

CLI Commands (Tool Usage)

Project Setup

• geronimo init — Scaffold a new ML project with SDK classes
  • --template realtime|batch|both — Choose project type
  • --framework sklearn|xgboost|lightgbm|pytorch — ML framework

Code Generation

• geronimo generate terraform — Generate AWS infrastructure (ECS Fargate, ALB, ECR)
• geronimo generate dockerfile — Create optimized Dockerfile with UV
• geronimo generate pipeline — Generate CI/CD pipelines (Azure DevOps)
• geronimo generate batch — Create Metaflow/Airflow job definitions
• geronimo generate all — Generate all artifacts at once

Validation

• geronimo validate — Check configuration against deployment rules

API Key Management

• geronimo keys create — Generate new API key with scopes
• geronimo keys list — List all managed keys
• geronimo keys revoke — Disable a key (keeps record)
• geronimo keys delete — Permanently remove a key

Monitoring

• geronimo monitor snapshot — Capture current model state
• geronimo monitor drift — Compare snapshots for drift

---

Installation

pip install geronimo

Quick Start

# Create a new project
geronimo init --name my-model --framework sklearn --template realtime
# Generate deployment artifacts
cd my-model
geronimo generate all
# Validate configuration
geronimo validate

License

Apache 2.0