AgentSkillsCN

ai-ml-engineer

针对模型选择、训练、评估、MLOps 以及安全部署的机器学习/AI 工程指导。适用于设计、实施或评审 ML/AI 功能、数据管道或模型治理方案时使用。

SKILL.md
--- frontmatter
name: ai-ml-engineer
description: Machine learning/AI engineering guidance for model selection, training, evaluation, MLOps, and safe deployment. Use when designing, implementing, or reviewing ML/AI features, pipelines, or model governance.

AI/ML Engineer

Scope

  • •Build and maintain ML/AI pipelines, models, and inference services.
  • •Align with security, privacy, and compliance requirements.

Workflow

  1. •Clarify objective, success metrics, and constraints.
  2. •Define data sources, labeling strategy, and data quality checks.
  3. •Select model approach and baseline; document rationale.
  4. •Implement training pipeline with reproducibility and versioning.
  5. •Evaluate with robust metrics and bias/fairness checks.
  6. •Package model with monitoring, drift detection, and rollback plan.

Deliverables

  • •Model card (purpose, data, metrics, limitations).
  • •Training/inference pipeline documentation.
  • •Monitoring and alerting plan.

Guardrails

  • •Avoid using sensitive data without explicit approval.
  • •Ensure deterministic training configs and seeded runs.
  • •Keep inference latency and cost constraints visible.