Join Snowflake for the virtual Goodyear ML Workshop on April 29th from 9:00 AM to 11:30 AM ET.
In this session, we will dive into how Snowflake simplifies the MLOps lifecycle. Key objectives include:
- Consolidation: Demonstrating how Snowflake serves as a unified platform for Goodyear’s ML ecosystem.
- Lifecycle Management: Showcasing the end-to-end journey from ad-hoc Python development to enterprise-scale deployment.
- The Snowflake Advantage: Highlighting robust security, governance, and key competitive differentiators compared to AWS.
AGENDA
9:00 – 9:30 AM | Setting the Foundation (30 min)
Welcome & Context Alignment (10 min)
- Recap Goodyear’s current state: AWS Sagemaker and Metaflow pipelines
- Workshop goals: show path from experimentation → simplified MLOps
Snowflake’s ML/AI Vision (20 min)
- Platform approach: unified data + ML + governance
- Architecture overview: Snowpark, Cortex Code, Model Registry, Feature Store, Monitoring, Experiments
- Competitive positioning vs AWS SageMaker / Metaflow (cost, data movement, governance)
- Security & governance built-in (role-based access, data masking, audit logs)
9:30 – 10:15 AM | Simplifying ML Pipelines with Snowpark (45 min)
From Ad-Hoc to Reusable: Snowpark ML (15 min)
- Write Python once, deploy anywhere (notebooks, tasks, stored procedures, UDFs)
- Feature engineering at scale with Snowpark DataFrames
- Model training without data movement
- Dependency and library management
Demo 1: End-to-End Forecasting Pipeline (30 min)
Use Case: Time-series forecasting (align with Goodyear’s forecasting models)
Walkthrough:
- Load historical data (manufacturing/sales/demand data)
- Feature engineering in Snowpark Python (window functions, aggregations)
- Train forecasting model (Prophet, XGBoost, or scikit-learn)
- Register model to Model Registry with versioning
- Deploy as UDF for batch predictions or real-time scoring
- Schedule with Snowflake Tasks for automated retraining
10:15 – 10:45 AM | Advances MLOps & Feature Store (30 min)
Enterprise MLOps Capabilities (15 min)
- Model Registry: versioning, tagging, deployment tracking
- Feature Store: centralized feature definitions, reusability, consistency
- CI/CD integration with Git (Snowflake Git Integration)
- Monitoring & observability (model performance tracking, drift detection)
Demo 2: Feature Store in Action (15 min)
Use Case: Manufacturing features for predictive maintenance
Walkthrough:
- Define reusable features (equipment metrics, production rates, quality indicators)
- Create feature views with refresh schedules
- Show feature discovery and lineage
- Use features across multiple models (forecasting, anomaly detection)
10:45 – 10:55 AM | BREAK
10:55 – 11:05 AM | Wrap-Up and Q&A
Don’t miss out, be sure to register on this page to confirm your participation!
Register Here