用多智能体框架把自然语言指令转为可验证的云部署，确保每步操作都可靠。
Forward-Deployed Full-Stack Engineering for Autonomous Cloud MLOps
- 通过带证据门控的多智能体协同生成和验证代码仓库与部署流程。
- 实验表明能阻止不合规流程，确保每次运行要么成功部署，要么留下可审计失败记录。
- 适合需要高可靠性、自动化运维的AI工程团队使用。
在各行业中，机器学习系统支持从预测、异常检测到优化与调度的应用，但其落地需协调应用开发、模型流水线、云基础设施、安全、部署、监控、重训练、恢复与回滚等环节。本文提出一种基于证据门控的多智能体框架，将自然语言的MLOps云工程任务转化为可验证的代码仓库与实际云部署。该框架结合图工程、循环工程与智能体约束工程，由有状态的图编排器协调多个专用智能体完成仓库生成、审查、执行、验证、发布与监控，并管理流程依赖、证据门限、重试上限、恢复路径与终止机制。关键生命周期转换仅在可验证执行或运行时证据支持下才触发。验证失败激活有限反思、修复与重验证；运行时出现故障、漂移、退化或策略违规时，可触发有限自适应、恢复或回滚。智能体约束工程通过受控能力与隔离环境限制仓库生成、审查与修复、制品执行及云操作。我们在Google Cloud Platform上实现该框架，并评估了仓库完整性、受控执行、证据门控转换、云发布与有限恢复能力。实验结果表明，该框架有效防止非支持的生命周期转换，并推动每次运行最终走向已验证的生产部署或可审计的终端失败。
原文摘要 · Abstract (English)
Across industries, machine-learning systems support applications ranging from prediction and anomaly detection to forecasting, optimization, and scheduling, yet operationalizing these systems requires coordinating application development, model pipelines, cloud infrastructure, security, deployment, monitoring, retraining, recovery, and rollback. We present an evidence-gated multi-agent framework for transforming a natural-language MLOps cloud engineering task into a verified repository and operational cloud deployment. The framework combines graph engineering, loop engineering, and agent harness engineering. A stateful Graph Orchestrator coordinates specialized agents for repository generation, review, execution, verification, release, and monitoring while governing workflow dependencies, evidence gates, retry bounds, recovery paths, and termination. Consequential lifecycle transitions proceed only when their required predicates are supported by verifiable execution or runtime evidence. Verification failures activate bounded reflection, repair, and re-verification, while runtime evidence of failure, drift, degradation, or policy violation can trigger bounded adaptation, recovery, or rollback. Agent harness engineering constrains repository generation, review, and repair, artifact execution, and cloud operations through controlled capabilities and isolated execution environments. We realize the framework on Google Cloud Platform and evaluate repository completeness, controlled execution, evidence-gated transitions, cloud promotion, and bounded recovery. Our experimental results show that the framework prevents unsupported lifecycle transitions and drives each run toward either a verified operational deployment or an auditable terminal failure.
Thank you to arXiv for use of its open access interoperability. PaperDance 不是 arXiv 官方产品；中文卡片由大模型生成，请以原文为准。