cursor-rules / rules

---
description: AWS SageMaker Development Standards
globs: 
alwaysApply: false
---
---
Description: AWS SageMaker Development Standards
Globs: **/sagemaker/**/*, **/ml/**/*, **/models/**/*, **/notebooks/**/*
---

# AWS SageMaker Development Standards
@base.mdc
@typescript.mdc

## SageMaker Project Structure
- Organize projects with clear separation of concerns
- Implement proper model training/inference code separation
- Create reproducible experiment environments
- Design modular processing pipelines
- Implement proper data preprocessing components
- Create appropriate model evaluation modules
- Design proper feature engineering pipelines
- Implement proper deployment configurations

## Model Development
- Implement standardized model interfaces
- Create proper model parameter configurations
- Design appropriate hyperparameter optimization
- Implement proper validation strategies
- Create appropriate model evaluation metrics
- Design proper model versioning
- Implement appropriate model documentation
- Create proper reproducibility controls

## Data Management
- Implement proper data versioning
- Create appropriate data preprocessing pipelines
- Design efficient data loading mechanisms
- Implement proper feature extraction
- Create appropriate data validation checks
- Design proper data labeling processes
- Implement efficient data augmentation
- Create appropriate data splitting strategies

## Training Jobs
- Configure appropriate training job parameters
- Implement proper distributed training
- Create appropriate checkpoint strategies
- Design efficient hyperparameter tuning jobs
- Implement proper early stopping
- Create appropriate resource configurations
- Design proper training metrics collection
- Implement appropriate experiment tracking

## Model Deployment
- Implement proper model serving configurations
- Create appropriate endpoint configurations
- Design efficient auto-scaling policies
- Implement proper model monitoring
- Create appropriate deployment strategies
- Design proper A/B testing mechanisms
- Implement efficient batch transform jobs
- Create appropriate multi-model endpoints when applicable

## SageMaker Pipelines
- Design proper ML workflow pipelines
- Implement appropriate pipeline components
- Create modular pipeline steps
- Design proper pipeline parameters
- Implement appropriate condition steps
- Create proper pipeline versioning
- Design effective data processing steps
- Implement proper model evaluation steps

## MLOps Practices
- Implement proper CI/CD for ML models
- Create appropriate model registries
- Design effective model approval workflows
- Implement proper model lineage tracking
- Create appropriate model artifact management
- Design proper model rollback strategies
- Implement efficient model monitoring
- Create proper automated retraining workflows

## Model Monitoring
- Configure appropriate model monitors
- Implement proper data quality monitoring
- Create appropriate model quality monitoring
- Design effective model bias monitoring
- Implement proper model explainability metrics
- Create appropriate alerting thresholds
- Design proper baseline constraints
- Implement efficient monitoring schedules

## Notebooks and Experimentation
- Implement structured notebook templates
- Create reproducible notebook environments
- Design proper experiment tracking
- Implement efficient notebook parameterization
- Create appropriate visualization standards
- Design modular notebook components
- Implement proper notebook version control
- Create appropriate notebook documentation

## Cost Optimization
- Implement right-sized instance selection
- Create appropriate spot instance usage
- Design efficient resource cleanup
- Implement proper auto-scaling configurations
- Create appropriate storage lifecycle policies
- Design cost-efficient training strategies
- Implement proper resource tagging
- Create appropriate cost allocation tracking

## Security Best Practices
- Implement proper IAM roles with least privilege
- Create appropriate VPC configurations
- Design proper encryption at rest and in transit
- Implement secure network configurations
- Create appropriate model artifact encryption
- Design proper data access controls
- Implement proper logging and monitoring
- Create appropriate security group configurations

## SageMaker Studio
- Implement proper user management
- Create appropriate domain configurations
- Design proper app lifecycle management
- Implement efficient notebook sharing
- Create appropriate resource isolation
- Design proper environment configurations
- Implement appropriate custom image management
- Create proper collaboration workflows

## SageMaker Feature Store
- Design proper feature group definitions
- Implement appropriate feature engineering pipelines
- Create efficient feature retrieval patterns
- Design proper online/offline store usage
- Implement appropriate feature versioning
- Create proper feature lineage tracking
- Design efficient feature transformation
- Implement appropriate feature validation

## Model Inference
- Implement efficient serialization/deserialization
- Create appropriate preprocessing/postprocessing
- Design proper batch vs. real-time inference patterns
- Implement efficient model loading
- Create appropriate inference pipeline components
- Design proper multi-model serving strategies
- Implement appropriate model performance profiling
- Create proper model optimization (ONNX, etc.)

## Integration with Other AWS Services
- Implement proper S3 integration for data storage
- Create appropriate CloudWatch metrics and alarms
- Design effective Step Functions for orchestration
- Implement proper Lambda integration
- Create appropriate EventBridge integration
- Design effective integration with data lakes
- Implement proper integration with Glue for ETL
- Create appropriate API Gateway integration for model serving

More agent context in jimmypocock/cursor-rules

17 other files this repository gives its agents.