cloud mlops platform mapper

The problem

Choosing an MLOps stack often results in vendor lock-in or architectural debt because teams select tools based on brand features rather than workload constraints. Developers struggle to map complex model lifecycles to specific cloud services across AWS, Azure, and GCP without a formal decision framework.

What it does

• Analyzes workload requirements and team maturity to select the optimal MLOps control plane.
• Generates a structured CLOUD-MLOPS-ADR (Architecture Decision Record) for documentation and compliance.
• Identifies specific Terraform boundaries, providers, and identity requirements for the chosen stack.
• Produces a reversible implementation sequence to start migrations or new builds without high risk.
• Classifies all architectural findings as Verified, Inferred, or Assumed to highlight data gaps.

Frameworks & tools

AWS (SageMaker), Azure (Azure ML), GCP (Vertex AI), Terraform, and hybrid portability layers.

Why this beats prompting it yourself

Standard LLM prompts often hallucinate feature parity or give generic advice. This skill enforces an evidence contract that prevents assuming maturity from tool names and requires calibrated baselines for every recommendation, ensuring your ADR stands up to technical scrutiny.

Use cases

• Comparing Vertex AI vs SageMaker for a specific computer vision workload.
• Reviewing Terraform readiness for a new Azure ML deployment.
• Mapping a local MLflow workflow to a managed cloud control plane.
• Conducting a portability review for hybrid-cloud MLOps requirements.

Known limitations

Does not generate production infrastructure code or execute cloud deployments. Does not handle credentials or provide runtime validation of suggested architectures.