Описание
The AI Platform project builds a reliable versioning model for LLM agents as immutable deployment units. It supports controlled release workflows, evaluation-based promotion gates, and rollback patterns to maintain quality and consistency across environments.
Задачи
•Develop immutable versioning workflows for agent container images, prompt configuration, tool bindings, and policy associations as a single deployment unit
•Design and maintain CI/CD pipelines with evaluation gates for AI systems before production promotion
•Implement agent promotion workflows across development, QA, and production environments
•Support rollback patterns for agent versions and validate restored versions
•Manage AWS ECR image practices, including ARM64 tagging, lifecycle policies, and image signing for agent immutability
•Integrate agent versions with a registry and maintain metadata such as evaluation results and promotion status
•Collaborate on endpoint management patterns for default and named runtime endpoints
Требования
•5+ Years of MLOps or DevOps engineering experience with AI or ML systems
•Experience with versioning and lifecycle management for LLM agents or ML models
•Experience with immutable artifact versioning at enterprise scale using ECR or model registries
•Experience designing CI/CD pipelines with quality gates for AI systems
•Hands-on experience with AWS AgentCore Runtime versioning
•Experience with agent evaluation frameworks such as Langfuse, LangSmith, or PromptLayer
•Experience with multi-environment agent promotion patterns
•Experience with agent version comparison and regression detection
Условия
No conditions specified
