Specialist - Architecture

1 Feature Engineering Pipeline Management

Design and scale feature pipelines using Snowpark PythonSQL to transform raw data

into productionready ML features

Implement and manage the Snowflake Feature Store as a centralized governed repository

for batch training and lowlatency online inference

Optimize data ingestion and processing costs using Snowflakes elastic compute

multicluster warehouses and search optimization services

2 Model Training Orchestration

Establish scalable ML training infrastructure using Snowflake Notebooks and

Container Runtimes CPUGPU instances without data egress

Orchestrate endtoend ML workflows and retraining schedules using Snowflake Tasks

and Streams

Integrate opensource frameworks ScikitLearn PyTorch XGBoost into Snowflake

ecosystem via Snowpark ML

3 Model Deployment Serving

Manage the Snowflake Model Registry for cataloging versioning metadata logging and

lifecycle governance Development Staging Production

Deploy models for batch and realtime inference using UserDefined Functions UDFs

or containerized services

Integrate LLMs and GenAI applications using Snowflake Cortex AI functions

4 MLOps Platform Setup Operations

Design and build endtoend MLOps platform including CICD pipelines model registry

experiment tracking and feature store

Implement automated model training validation deployment and monitoring workflows

Establish reusable ML pipeline templates and accelerators for development teams

5 CICD Automation Infrastructure

Build automated CICD pipelines using Terraform for model testing validation and promotion

Implement InfrastructureasCode IaC using Terraform to provision Snowflake

resources securely and repeatably

Enforce data and model governance through Snowflakes native security

rowlevel security data masking RBAC

6 Monitoring Observability

Deploy ML monitoring frameworks to track model performance data drift

and prediction latency

Design automated retraining loops triggered by accuracy drops or data distribution shifts

Build operational dashboards using Streamlit in Snowflake for realtime model health visibility

7 Snowflake ML Standards Development

Define and enforce ML engineering standards patterns and best practices within Snowflake

Develop Snowflakenative ML workflows for feature engineering training and inference

Manage Snowflake computewarehouse configurations optimized for ML workloads

8 ML Development Assistance

Collaborate with data scientists to productionize models from prototype to

productiongrade code

Build shared libraries utilities and SDKs to accelerate model development

Conduct code reviews and enforce codingtesting standards for ML codebases

9 Governance Documentation

Establish model versioning lineage tracking and reproducibility standards

Document platform architecture runbooks and onboarding guides

Ensure compliance with RBCs data governance and security policies