There is no single, industry-standard “AI engineer certification” in 2026. Microsoft’s credential that actually carried the name, Azure AI Engineer Associate (exam AI-102), retired on June 30, 2026, so the closest real options are proctored exams tied to the platform you build on: Microsoft AI-103, AWS’s Generative AI Developer Professional and Machine Learning Engineer Associate, Google Cloud’s Professional Machine Learning Engineer, NVIDIA’s NCA-GENL, the Databricks Generative AI Engineer Associate and Anthropic’s Claude Certified Developer: Foundations.
They cost $125 to $300 (AWS’s machine learning engineer exam is $75 while its new version is in beta) and take one to three hours. Below are all seven side by side, what each one proves, how to pick by the stack you use, and where to practice for free before you book.
AI engineer certifications at a glance
| Official “AI engineer” certification | None current from a major vendor. Microsoft’s Azure AI Engineer Associate (AI-102) retired June 30, 2026 |
| What replaced AI-102 | AI-103, Azure AI Apps and Agents Developer Associate. Microsoft lists its role as “AI Engineer” and “Developer” |
| Real options | 7 proctored exams from Microsoft, AWS (2), Google Cloud, NVIDIA, Databricks and Anthropic |
| Price range (US) | $125 to $300; the AWS ML Engineer Associate beta (MLA-C02) is $75 for now |
| Shortest exam | NVIDIA NCA-GENL: 1 hour, $125, entry level |
| Most senior | AWS Generative AI Developer Professional ($300, 180 minutes) and Google Professional ML Engineer ($200, 2 hours) |
| How long they last | Microsoft 1 year and Claude 12 months (both with free renewal); Google Professional, NVIDIA and Databricks 2 years; AWS 3 years |
| Open to everyone? | All except the Claude exams, which are for Claude Partner Network members only |
| Free practice questions here | AI-103, MLA-C02, NCA-GENL and CCDV-F |
Checked against each issuer’s own certification pages and exam guides (Microsoft Learn, AWS Certification, Google Cloud, NVIDIA, Databricks and Anthropic’s Partner Academy) on October 9, 2026.
The 7 AI engineer certifications compared
Every option below is a proctored exam: you book a slot and answer under a proctor’s watch, online or at a test center. Prices are US list prices before tax.
| Certification | Issuer | Format | Price | Length | What it proves |
|---|---|---|---|---|---|
| Azure AI Apps and Agents Developer Associate (AI-103) | Microsoft | Proctored exam (Pearson VUE) | $165 | 120 minutes | You can build, deploy and manage generative AI apps and agents on Azure with Microsoft Foundry and Python |
| AWS Certified Generative AI Developer – Professional (AIP-C01) | AWS | Proctored exam (Pearson VUE) | $300 | 180 minutes, 75 questions | You can take generative AI apps on AWS, such as Amazon Bedrock, from proof of concept to production with security and cost control |
| AWS Certified Machine Learning Engineer – Associate (MLA-C02, in beta) | AWS | Proctored exam (Pearson VUE) | $75 beta (MLA-C01 is $150) | 170 minutes, 85 questions (beta) | You can prepare data, build, deploy and monitor ML and foundation model workloads on SageMaker AI and Bedrock |
| Google Cloud Professional Machine Learning Engineer | Google Cloud | Proctored exam (online or test center) | $200 | 2 hours, 50–60 questions | You can build, productionize, serve and monitor ML and generative AI solutions on Google Cloud |
| NVIDIA-Certified Associate: Generative AI LLMs (NCA-GENL) | NVIDIA | Proctored exam (Certiverse account) | $125 | 1 hour, 50–60 questions | Entry-level knowledge of developing, integrating and maintaining LLM applications with NVIDIA solutions |
| Databricks Certified Generative AI Engineer Associate | Databricks | Proctored exam (online or test center) | $200 | 90 minutes, 45 scored questions | You can build and deploy RAG applications and LLM chains with Databricks tools |
| Claude Certified Developer: Foundations (CCDV-F) | Anthropic | Proctored exam (Pearson VUE); partners only | $125 | 120 minutes, 53 questions | You can build, integrate and ship Claude apps, agents and workflows |
Want the full cost picture, including retakes and renewals? Our guide to what AI certifications cost and the AI certification cost calculator add it up for you.
Certification or certificate? Know what you are buying
Search for “AI engineer certificate” and you’ll also find online professional certificates, bootcamps and university courses. Those are course certificates: you earn them by finishing lessons and projects, and nobody checks under exam conditions that you did the work yourself. They can be a good way to learn. A certification is different: a timed, proctored exam set by the company whose platform you’ll use at work.
You’ll also see vendor-neutral programs that sell a “certified AI engineer” title. Before paying, check two things: does it end in a proctored exam, and do the employers you want to work for mention it in their job postings? If they ask for Azure, AWS or Google Cloud, the vendor exam will carry more weight.
Pick an AI engineer certification by stack
The best AI engineer certification is the one for the platform you build on, or the one named in the job postings you’re targeting.
| If you build on… | Start with | Next step |
|---|---|---|
| Azure and Microsoft Foundry | AI-103 | AI-300 for MLOps, or the expert-level Multi-Agent AI Solutions Expert (AI-500), which requires AI-103 |
| AWS, training and deploying models | ML Engineer Associate (MLA-C02) | Generative AI Developer Professional |
| AWS, generative AI apps on Amazon Bedrock | Generative AI Developer Professional (AIP-C01) | AWS suggests 2+ years building on AWS or open source and 1 year of hands-on generative AI first |
| Google Cloud | Professional Machine Learning Engineer | Google recommends 3+ years of industry experience, including 1+ year on Google Cloud |
| NVIDIA GPUs and open models | NCA-GENL | NVIDIA-Certified Professional: Generative AI LLMs (NCP-GENL), $200, 120 minutes |
| Databricks | Generative AI Engineer Associate | Databricks recommends 6+ months of hands-on generative AI work |
| Claude, at a Claude partner company | CCDV-F | The Claude architect exams; see the Claude certification guide |
Not tied to a platform yet? Start with the cloud your target employers use. If you need a cheaper first step, NCA-GENL is entry level at $125 and one hour, and our list of AI certifications for beginners covers the foundational exams that sit below these. The generative AI certifications guide compares every gen AI credential by level.
The 7 options in detail
Microsoft AI-103: Azure AI Apps and Agents Developer Associate
AI-103 is Microsoft’s replacement for AI-102, and its certification page lists the job role as “AI Engineer” and “Developer.” It costs $165 in the US (prices vary by country), runs 120 minutes, and you need 700 to pass. Microsoft expects experience building apps in Python. The skills outline (as of April 16, 2026) puts the most weight on implementing generative AI and agentic solutions (30–35%), then planning and managing an Azure AI solution (25–30%), with computer vision, text analysis and information extraction at 10–15% each. Like other Microsoft associate certifications, it lasts one year and renews free online.
Go deeper: AI-103 study guide, AI-103 exam cost, how to pass AI-103, is AI-103 worth it and AI-102 vs AI-103.
AWS Certified Generative AI Developer – Professional (AIP-C01)
This is AWS’s professional-level AI exam: $300, 180 minutes, 75 questions (65 scored) and a passing score of 750. AWS aims it at developers with 2+ years of building production applications and 1 year of hands-on generative AI work. Its largest domain is foundation model integration, data management and compliance (31%), followed by implementation and integration (26%) and AI safety, security and governance (20%). It’s valid for 3 years. See the AIP-C01 study guide and AIP-C01 exam cost.
AWS Certified Machine Learning Engineer – Associate (MLA-C02)
The ML Engineer Associate is changing versions. The last day to take MLA-C01 in English was September 28, 2026, and the new MLA-C02 has been in beta since September 29: $75, 170 minutes, 85 questions, English only. AWS lists ML engineer, MLOps engineer and LLMOps engineer among the candidate roles and asks for at least a year with SageMaker AI, Bedrock and other ML services. The four domains stay the same as MLA-C01, but MLA-C02 adds foundation models, Bedrock and RAG. It’s valid for 3 years.
Go deeper: what changes in MLA-C02, MLA-C02 study guide, ML Engineer exam cost and is it worth it.
Google Cloud Professional Machine Learning Engineer
Google’s exam costs $200 plus tax, runs 2 hours with 50–60 multiple choice and multiple select questions, and is offered in English and Japanese, online or at a test center. There’s no prerequisite, but Google recommends 3+ years of industry experience including at least a year designing and managing solutions on Google Cloud. The exam guide dated June 1, 2026 reflects the move from Vertex AI to Gemini Enterprise Agent Platform. Google says it doesn’t directly test coding, though you should read Python and SQL well enough to follow code snippets. Professional certifications are valid for 2 years. We don’t have a Google ML Engineer guide or practice questions yet; start with Google’s exam guide and sample questions on the certification page.
NVIDIA NCA-GENL: Generative AI LLMs Associate
NVIDIA calls NCA-GENL an entry-level credential. It’s $125, 1 hour, 50–60 multiple choice questions in English, taken through a Certiverse account. Results are pass or fail with no score. The topics: core machine learning and AI knowledge (30%), software development (24%), experimentation (22%), data analysis (14%) and trustworthy AI (10%). It’s valid for 2 years, and you renew by retaking the exam. NVIDIA’s next step for engineers is the Professional NCP-GENL ($200, 120 minutes, 60–70 questions, with 2–3 years of LLM experience recommended).
Go deeper: NCA-GENL study guide, NCA-GENL exam cost and is NCA-GENL worth it.
Databricks Certified Generative AI Engineer Associate
Databricks’ exam checks that you can design and implement LLM-enabled solutions on its platform, including AI Search for semantic similarity search, Model Serving, MLflow and Unity Catalog. It’s $200 with 45 scored questions in 90 minutes, online or at a test center, in English, Japanese, Brazilian Portuguese or Korean. Application development is the biggest section (30%), then assembling and deploying apps (22%). There’s no prerequisite, but Databricks recommends 6+ months of hands-on generative AI work and says there are no free retakes. It’s valid for 2 years. Details are on the Databricks certification page.
Claude Certified Developer: Foundations (CCDV-F)
Anthropic’s developer exam is for engineers who build and ship Claude apps, agents and workflows. It costs $125, has 53 questions in 120 minutes, and is delivered by Pearson VUE in English only. The biggest blueprint area is applications and integration (33.1%), followed by model selection and optimization (16.8%), agents and workflows (14.7%), prompt and context engineering (11.0%) and tools and MCPs (10.6%). The catch: Claude certifications are only open to people at Claude Partner Network organizations, registering with a company email. A pass lasts 12 months, with a free renewal assessment.
Go deeper: CCDV-F study guide, how to pass CCDV-F and all four Claude certifications.
Other engineer-level AI exams worth knowing
- Microsoft AI-300, Machine Learning Operations Engineer Associate: MLOps and GenAIOps on Azure Machine Learning and Microsoft Foundry; $165, 120 minutes, English only. It replaced the retired DP-100. See the AI-300 study guide and is AI-300 worth it.
- Microsoft Multi-Agent AI Solutions Expert (AI-500): expert level for production multi-agent systems; you must hold the AI-103 certification first.
- Microsoft AI-200, Azure AI Cloud Developer Associate: for developers adding AI features to Azure apps. See the AI-200 study guide.
- NVIDIA NCP-GENL: the professional-level LLM exam covering fine-tuning, model optimization and GPU acceleration.
For every AI exam we track, with vendor-checked facts, see the AI certification index. For a ranked shortlist across roles, see the best AI certifications.
Which AI engineer certification should you get first?
- You want a job at a company that runs on one cloud: take that cloud’s exam. AI-103 for Azure, MLA-C02 or AIP-C01 for AWS, Professional ML Engineer for Google Cloud.
- You build generative AI apps and agents: AI-103 is the most direct match at $165; AIP-C01 is the senior AWS option at $300.
- You train, deploy and monitor models: MLA-C02 on AWS, Google’s Professional ML Engineer, or AI-300 on Azure.
- You want the cheapest credible start: NCA-GENL at $125 and one hour, or the MLA-C02 beta at $75 while it lasts.
- Your company is a Claude partner: CCDV-F, then the architect exams.
Wondering whether any of this pays? Our honest look at whether AI certifications increase salary covers what a certification can and can’t do for your career.
Practice before you book
Our free practice questions are original, written from each vendor’s published exam guide, never copied from a real exam. Four of the seven exams on this page have a free bank here:
- AI-103 practice questions
- MLA-C02 practice questions
- NCA-GENL practice questions
- CCDV-F practice questions
We don’t have practice questions for AWS AIP-C01, Google’s Professional ML Engineer or the Databricks exam yet. Use the official sample questions and practice sets on each vendor’s certification page, and skip anyone selling “real exam questions”: dumps break the vendors’ exam agreements and can cost you the certification. Browse every free practice exam on AI Exam Prep →
HOW TO // AI is not affiliated with or endorsed by Microsoft, Amazon Web Services, Google, NVIDIA, Databricks or Anthropic. Certification names are trademarks of their owners and are used here descriptively. Prices and exam details change, so check the official certification page before you book.
Get AI Cert Watch: an email when AI certification exams change
We check the official pages of 80+ AI exams every day. When a price, version, format or retirement date changes, you’ll get a short email. At most one a week, plus the HOW TO // AI newsletter on Thursdays.
Unsubscribe anytime. Privacy policy.
Related guides
- AI-103 Study Guide & Cheat Sheet (Azure AI Apps & Agents Developer)
- AWS Generative AI Developer Professional (AIP-C01) Study Guide
- MLA-C02 Exam Guide (ME1-C02): What Changed From MLA-C01
- NVIDIA NCA-GENL (Gen AI & LLMs) Study Guide & Cheat Sheet
- Claude Developer Study Guide & Cheat Sheet (CCDV-F)
Frequently asked questions
Is there an official AI engineer certification?
Not from a major vendor as of October 2026. Microsoft's Azure AI Engineer Associate (AI-102) retired on June 30, 2026, and the closest options now are platform exams such as Microsoft AI-103, AWS Generative AI Developer Professional, Google Cloud Professional Machine Learning Engineer and NVIDIA NCA-GENL.
What replaced the Azure AI Engineer Associate certification?
Microsoft recommends AI-103, which earns the Azure AI Apps and Agents Developer Associate certification. It costs $165 in the US, runs 120 minutes, and focuses on generative AI apps and agents built with Microsoft Foundry and Python.
What is the best AI engineer certification?
The one for the platform you build on. AI-103 for Azure, MLA-C02 or AIP-C01 for AWS, the Professional Machine Learning Engineer for Google Cloud, NCA-GENL for NVIDIA, the Generative AI Engineer Associate for Databricks, and CCDV-F if your company is a Claude partner.
How much does an AI engineer certification cost?
Most cost $125 to $300 in the US. NVIDIA NCA-GENL and Claude CCDV-F are $125, AI-103 is $165, Google and Databricks are $200, and AWS Generative AI Developer Professional is $300. The AWS ML Engineer Associate beta (MLA-C02) is $75 for now.
Which AI engineer certification is best for beginners?
NVIDIA NCA-GENL, which NVIDIA calls an entry-level credential: $125, one hour and no required experience beyond a basic understanding of generative AI and LLMs. The others expect hands-on building experience, from a year with AWS ML services to Google's recommended 3+ years.
Do AI engineer certifications expire?
Yes. Microsoft associate certifications last one year and Claude certifications 12 months, both with a free renewal. Google Professional, NVIDIA and Databricks certifications last 2 years, and AWS certifications last 3 years.
Do I need to know Python for an AI engineer certification?
For most, yes. Microsoft expects Python app development experience for AI-103, Google says you should read Python and SQL snippets, and Anthropic recommends Python or TypeScript for CCDV-F.
Can anyone take the Claude developer certification?
No. Claude certifications, including CCDV-F, are only open to people at Claude Partner Network organizations, who register with a company email address.
Is an online AI engineering certificate the same as a certification?
No. A course certificate shows you finished a course; a certification means you passed a proctored exam set by the vendor. If the job postings you want name Azure, AWS or Google Cloud, that vendor's exam is the stronger proof.
