senior-ml-engineer_skill

Overview

This skill packages world-class senior ML engineering expertise for productionizing ML models, MLOps, and building scalable ML systems. It focuses on end-to-end delivery: architecture, deployment, monitoring, and operational excellence for both conventional ML and LLM-driven systems. Use it to design reliable, high-performance ML platforms and lead implementation across teams.

How this skill works

The skill inspects your production requirements, infrastructure constraints, and model characteristics to recommend architecture patterns, deployment pipelines, and monitoring strategies. It translates requirements into concrete artifacts: CI/CD flows, serving/topology design, feature-store integration, retraining triggers, and observability plans. It also provides practical commands, scripts, and configuration recommendations for common stacks (Kubernetes, Docker, cloud providers, TensorFlow/PyTorch, and LLM tooling).

When to use it

• Deploying or scaling ML models from prototype to production
• Designing MLOps pipelines, automated retraining, and feature stores
• Integrating LLMs, RAG systems, or fine-tuning workflows into products
• Implementing monitoring, drift detection, and model observability
• Optimizing inference latency, throughput, and cost for large-scale systems

Best practices

• Design for idempotent, observable pipelines with automated tests and CI/CD
• Use feature stores and versioned data to ensure reproducible training and serving
• Implement gradual rollouts (feature flags, canary/A/B tests) and automated rollback
• Monitor both data and model signals (drift, performance, latency) and trigger retraining
• Prioritize security: auth, encryption, PII handling, and regular audits

Example use cases

• Build a low-latency model serving infrastructure on Kubernetes with autoscaling and caching
• Create a retraining pipeline that detects data drift and triggers scheduled retrains with validation gates
• Integrate an LLM with a RAG system for product search, including vector DB and relevance monitoring
• Design cost-optimized inference for high-throughput APIs with batching and load balancing
• Establish team processes for code reviews, TDD, and mentoring to raise engineering quality