Understand the certification
认证概述
Google Cloud Certified - Professional Machine Learning Engineer (PMLE) 验证您设计、构建和产品化 ML 模型的能力，包括使用 Vertex AI 等 GCP 服务解决实际业务问题。
考试领域
- Architect low-code ML solutions（设计低代码 ML 解决方案）— 12%
- Collaborate within and across teams（团队内外协作）— 14%
- Scale prototypes into ML models（将原型扩展为 ML 模型）— 18%
- Serve and scale models（模型服务与扩缩）— 19%
- Automate and orchestrate ML pipelines（ML 管道自动化与编排）— 21%
- Monitor ML solutions（ML 解决方案监控）— 16%
适合人群
- 机器学习工程师和 MLOps 工程师
- 数据科学家希望提升工程化能力
- AI 平台架构师
- 希望验证 GCP ML 能力的 IT 专业人员
Exam at a glance
- Exam code
- PMLE
- Provider
- GCP
- Format
- Single choice / Multiple select / 案例分析题
- Language
- English / 日本語 / 한국어
- Validity
- 2 years
- Exam fee
- $200
Skills in focus
Learning outcomes
- 获得 Google Cloud 官方专业级 ML 工程师认证
- 掌握 Vertex AI 全栈 ML 平台的使用
- 具备端到端 ML 模型的训练、部署和监控能力
- 能够设计和实施 MLOps 流水线
- 提升 ML 工程师岗位竞争力
Who this is for
- 机器学习工程师和 MLOps 工程师
- 数据科学家希望提升工程化和产品化能力
- AI 平台架构师
- 后端工程师希望转型 ML 方向
- 希望验证 GCP ML 全栈能力的 IT 专业人员
Before you start
- 3 年以上 ML 或相关经验（含 1 年以上 GCP 实践）
- 熟悉 Python 编程和 ML 框架（TensorFlow/PyTorch）
- 了解机器学习基本概念和算法
- 建议有模型训练和部署的实际经验
What the exam tests
12%Architect Low-Code ML Solutions
Architect Low-Code ML Solutions
14%Collaborate Within and Across Teams
Collaborate Within and Across Teams
18%Scale Prototypes into ML Models
Scale Prototypes into ML Models
19%Serve and Scale Models
Serve and Scale Models
21%Automate and Orchestrate ML Pipelines
Automate and Orchestrate ML Pipelines
16%Monitor ML Solutions
Monitor ML Solutions
Your learning path
8 chapters
Plan your preparation
- Learning hours
- 60h
- With relevant experience
- 8-10 weeks
- Starting from scratch
- 12-16 weeks
- Daily study
- 1-2h
Included in your preparation
- 8 个系统化学习章节，覆盖考试 6 大领域
- 963 精选练习题，每题附详细解析
- 仿真模拟考试（50 题/120 分钟）
- 错题本 + 智能复习，薄弱环节针对性突破
- 学习进度可视化追踪
- 同步 Google Cloud 官方最新考纲
Preparation tips
- 60 题 120 分钟，平均每题 2 分钟，合理分配时间
- 及格分 70/100，不确定的题先标记跳过，回头再做
- 排除法非常有用 — 先排掉明显错误的选项，剩下的再分析
- 多选题会告诉你选几个，仔细看题目要求
Common mistakes
- 没有读完所有选项就选答案 — 题目经常有"最佳"答案和"正确但不最佳"的干扰项
- 备考只刷题不理解原理 — 考试场景题需要理解底层概念
- 忽略时间管理 — 在难题上卡太久，导致后面简单题没时间做
Before you register
How difficult is the GCP Professional ML Engineer exam?
PMLE is one of the most difficult GCP certifications. The exam tests not only GCP ML services (Vertex AI) but also dives deep into ML theory, model optimization, and MLOps practices. It is recommended to have 4-6 months of systematic preparation time.
What is the difference between PMLE and AWS MLA-C01?
PMLE focuses more on the Vertex AI platform and GCP-native ML services, with a high proportion of MLOps and model monitoring in the exam. AWS MLA focuses on the SageMaker ecosystem. Both test similar core ML concepts, but use different platform tools.
Do I need to master TensorFlow programming?
You do not need to write complete code, but you need to understand core concepts of TensorFlow and Keras, such as model architecture, training loops, and distributed training strategies. The exam focuses more on architecture design and service selection.
How long is the PMLE certification valid?
The GCP Professional ML Engineer certification is valid for 2 years. The exam fee is $200 USD, duration is 120 minutes, and the passing score is approximately 70%.
Does Vertex AI have a high weight in the exam?
Yes, Vertex AI is the most central service in the exam, covering multiple components including AutoML, Custom Training, Prediction, Feature Store, Model Monitoring, and Pipelines. Almost every domain involves Vertex AI.
Learner experiences
Loading reviews…
Start with the first chapter.
2 free chapters · Full preparation from $39