eraneos

eraneos / snowflake-mlops

Public

Open-source reference monorepo for end-to-end MLOps on Snowflake, centered around Snowflake-native MLOps tooling (Feature Store, Model Registry, Tasks) in a hub-spoke layout. Questions, issues, or ideas about Snowflake ML in general are warmly welcomed.

eraneos.github.iosnowflake-mlops hub-and-spoke mlops mlops-environment mlops-platform mlops-project
19
0
100% credibility
Found May 19, 2026 at 19 stars -- GitGems finds repos before they trend. Get early access to the next one.
Sign Up Free
AI Analysis
Python
AI Summary

This is an open-source reference project showing how to build complete machine learning pipelines using Snowflake's built-in tools. The example focuses on predicting capacity at parcel pickup/dropoff locations (like parcel lockers and shops) in a logistics network. It demonstrates how to organize work across teams using a hub-and-spoke approach, how to store and reuse features for ML models, how to automate model training on schedules, and how to generate predictions that can be evaluated against real outcomes. All the data is synthetic and generated locally so anyone can run through the entire workflow without needing real data sources.

How It Works

1
🔍 Discover a Real-World MLOps Example

You find this open-source project while looking for how companies actually do machine learning on Snowflake in production.

2
🏗️ Set Up Your Foundation

You create the shared platform infrastructure that your whole team will use - databases, storage areas, and access permissions.

3
📦 Fill Your Database with Realistic Data

You populate your system with synthetic parcel delivery data that looks just like real logistics operations in Berlin.

4
🎯 Build Your Feature Store

You define the patterns your model will look for - historical delivery trends, nearby locations, and time-based patterns.

5
🏋️ Train Your Prediction Model

You deploy an automated training pipeline that learns from your data and gets smarter over time.

6
Generate Capacity Predictions
⏰
Scheduled Predictions

Predictions run automatically on a schedule so you always have fresh forecasts ready

🖥️
On-Demand Predictions

You trigger predictions whenever you need them, for any date you specify

7
📊 See How Well Your Model Performs

You compare your predictions against what actually happened and get clear metrics on accuracy.

🎉 Your ML Pipeline Is Live

You've built a complete, production-ready machine learning system that predicts parcel pickup point capacity - ready to share with your team.

Sign up to see the full architecture

6 more

Sign Up Free

Star Growth

See how this repo grew from 19 to 19 stars Sign Up Free
Repurpose This Repo

Repurpose is a Pro feature

Generate ready-to-use prompts for X threads, LinkedIn posts, blog posts, YouTube scripts, and more -- with full repo context baked in.

Unlock Repurpose
AI-Generated Review

What is snowflake-mlops?

This is an open-source reference implementation showing how to do production-grade MLOps entirely within Snowflake. Instead of stitching together external tools, it wires together Snowflake's native Feature Store, Model Registry, and Tasks DAGs into a complete workflow. The example project predicts parcel pickup/drop-off point capacity -- think predicting how full lockers and shops will be on any given day. Everything runs in Python, uses `uv` for package management, and generates synthetic logistics data locally so you can run the full pipeline without real upstream sources.

Why is it gaining traction?

The hook is "Snowflake-native MLOps without the glue code." Most ML platforms require juggling separate services for orchestration, feature management, and model versioning. This repo shows you can do it all with Snowpark, SQL-based task graphs, and the built-in registry. The hub-spoke architecture also gives teams a concrete pattern for organizing multi-team ML platforms -- central infrastructure plus team-owned projects. The extensive documentation with tutorials, concepts, and guides makes the learning curve gentler than rolling your own.

Who should use this?

Platform engineers building internal ML infrastructure on Snowflake will find the most value here. Data scientists who want to understand how to structure training and inference pipelines that live entirely in the warehouse. Teams evaluating whether Snowflake's native MLOps primitives are mature enough for production use. Not ideal for quick prototyping -- this is a reference architecture, not a plug-and-play library.

Verdict

With 19 stars and a 1.0% credibility score, this is a niche, early-stage project from a consulting firm. The documentation is genuinely thorough and the patterns are well-reasoned, but test coverage and community activity are minimal. Treat it as architectural inspiration rather than production-ready tooling. If you're committed to Snowflake-native ML and want a head start on the wiring, study the patterns; don't ship the code unchanged.

Sign up to read the full AI review Sign Up Free

Similar repos coming soon.