Model registry version mock in Python

A simple in-memory model registry that stores model versions with metadata and supports version listing and latest retrieval.

Easy Python 3.9+ Aug 9, 2026 ML engineering pipelines 13 views 0 copies

ml-engineering model-registry versioning data-structures

Python code

32 lines

Python 3.9+

class ModelRegistry:
    def __init__(self):
        self.models = {}

    def register(self, name, version, model_type, metrics=None):
        if name not in self.models:
            self.models[name] = []
        entry = {
            "version": version,
            "model_type": model_type,
            "metrics": metrics or {}
        }
        self.models[name].append(entry)
        return f"Registered {model_type} '{name}' version {version}"

    def get_latest(self, name):
        if name not in self.models or not self.models[name]:
            return None
        return self.models[name][-1]

    def list_versions(self, name):
        if name not in self.models:
            return []
        return [entry["version"] for entry in self.models[name]]


if __name__ == "__main__":
    registry = ModelRegistry()
    print(registry.register("sentiment-model", "1.0.0", "LogisticRegression", {"accuracy": 0.87}))
    print(registry.register("sentiment-model", "1.1.0", "RandomForest", {"accuracy": 0.91}))
    print("Versions:", registry.list_versions("sentiment-model"))
    print("Latest:", registry.get_latest("sentiment-model"))

Output

stdout

Registered LogisticRegression 'sentiment-model' version 1.0.0
Registered RandomForest 'sentiment-model' version 1.1.0
Versions: ['1.0.0', '1.1.0']
Latest: {'version': '1.1.0', 'model_type': 'RandomForest', 'metrics': {'accuracy': 0.91}}

How it works

The register method stores each model version as a new dict entry in a list keyed by model name. Using a list preserves insertion order, so the last appended version is treated as the latest. The get_latest method returns the final element of the list, while list_versions extracts just the version strings. Metrics default to an empty dict when not provided, keeping the data structure consistent.

Common mistakes

Forgetting to initialize the models list for a new model name before appending
Assuming versions arrive in chronological order instead of relying on insertion order
Returning the full entry instead of just the version string from `get_latest`

Variations

Use a dict keyed by (name, version) tuple for direct version lookups instead of a list
Add a semantic version comparison to `get_latest` so versions are sorted numerically rather than by insertion

Real-world use cases

Prototyping an ML experiment tracker before integrating with a full platform like MLflow.
Unit-testing model-serving code that needs a lightweight registry stub.
Managing rolling model deployments across environments in a CI/CD pipeline.

Model registry version mock in Python

Python code

Output

How it works

Common mistakes

Variations

Real-world use cases

More from ML engineering pipelines

Tutorials

Quizzes

Python code

Output

How it works

Common mistakes

Variations

Real-world use cases

More from ML engineering pipelines

Keep learning

Tutorials

Quizzes