Skip to the review
MenuClose

AI engineer certification: which exams exist and whether employers ask for them

Microsoft has retired its Azure AI Engineer exam and AWS has an updated ML Engineer exam in beta. We compared nine credentials and read 119 AI Engineer postings to see which, if any, employers name.

By roleBy AI Certification Reviews editorial teamPublished Updated Sources checked

We have a business connection with Torchlane. Money and recommendations

Digest:

  1. Our top pickTorchlane's AI FDE training, for working professionals who want to get into AI work or build on the expertise they already have: 3 to 6 client-style projects and a capstone, with project work assessed against seven published categories, and you can apply without coding experience 1,2. Its credentials are issued by Torchlane's own academy, are nonaccredited and are not a cloud vendor certification, and Torchlane says it is not a substitute for the engineering experience senior roles require 1,3. We have a business connection with Torchlane.
  2. 02The AI engineer exams you can sit on 23 October 2026 include Microsoft's AI-103, AWS's Machine Learning Engineer Associate and Generative AI Developer Professional, Google Cloud's Professional Machine Learning Engineer, NVIDIA's generative AI associate and Databricks' Generative AI Engineer Associate; among these vendor exams, published fees run from $75 (an AWS beta price) to $300, and Microsoft sets its fee by country 4,5,6,7,8,9.
  3. 03Microsoft has retired the Azure AI Engineer Associate and its renewal assessment; the AI-103 Azure AI Apps and Agents Developer Associate is tagged to the same AI Engineer role 10,4.
  4. 04In 119 AI Engineer postings from 31 employers, none named an AI exam; one listed certifications as a nice-to-have, and it was an Oracle one 11. Georgetown's CSET found little employer demand for AI certifications in 2010 to 2020 postings 12.
  5. 05Exams may count more at consultancies and cloud partners: AWS's services partner tiers set minimum numbers of certified staff 13.
Contents:

Short answer: choose the exam that matches the platform you will build on, and treat it as proof of platform knowledge rather than a hiring requirement, at least in the postings we read. As of 23 October 2026 the AI engineer exams you can sit include Microsoft's Azure AI Apps and Agents Developer Associate (AI-103), AWS's Machine Learning Engineer Associate and Generative AI Developer Professional, Google Cloud's Professional Machine Learning Engineer, NVIDIA's generative AI associate exam and Databricks' Generative AI Engineer Associate 4,5,6,7,8,9. Microsoft has retired the older Azure AI Engineer Associate 10. None of the 119 AI Engineer job postings we read named any of these exams.

How we checked. On 23 October 2026 we loaded each issuer's own exam page and describe every credential in its published terms. We did not sit any exam, and we do not score or rank them. For the job check we read the public job-board feeds (Greenhouse, Ashby and Lever) of 83 employers and took every posting whose title contains "AI Engineer" or "AI/ML Engineer" (Applied AI Engineer and Senior AI Engineer included): 119 postings from 31 employers, a non-random snapshot of mostly technology employers, not the whole market. Confirm fees with the issuer before you pay.

Which AI engineer certifications can you sit right now?

Every row is a credential an AI engineer might weigh, except our top pick, which is training.

Table 1 scrolls sideways inside its frame.

AI engineer credentials, as each issuer's page described them on 23 October 2026.
Credential and issuerHow it is assessedExperience asked or recommendedFee (checked 23 Oct 2026)Validity and renewal
Our top pick*: AI FDE training (Torchlane), which issues its own nonaccredited academy credentials (names provisional) 1,3Associate: every module quiz and a timed final exam. Professional: the final exam plus a human-reviewed capstone and mock interview. Projects (3 to 6, drawn from 12 simulated client engagements) are assessed against seven published categories, with a time-limited hidden test 3,1,2You can apply without coding experience; Torchlane says it is not a substitute for the engineering experience senior roles require 3,1Given in writing during admissions; no price is published 3Not stated; 90-day training goal, 120 days of program access, and a public verification link for issued credentials 3
Azure AI Apps and Agents Developer Associate (AI-103), Microsoft 4Proctored, 120 minutes; Azure AI solutions, generative AI and agents, vision, text analysis, information extraction 4Python app development; familiarity with generative AI and Azure 4Set by country or region 4Renews yearly through a free, unproctored, open-book assessment 14
Machine Learning Engineer, Associate, AWS 5Updated MLA-C02 in beta, English only: 85 questions, 170 minutes (beta exams include extra unscored items), adds Bedrock, RAG and agentic AI. Previous MLA-C01: 65 questions, 130 minutes 51 or more years with SageMaker AI, Bedrock and other ML engineering services 5$75 beta (MLA-C02); $150 (MLA-C01, no longer offered in English) 53 years; general availability dates not set 5
Generative AI Developer, Professional, AWS 675 questions, 180 minutes; test center or online proctored 62 or more years building production applications on AWS or open source, plus 1 year of hands-on generative AI 6$300 6Not stated on the exam page
Professional Machine Learning Engineer, Google Cloud 72 hours, 50 to 60 multiple-choice and multiple-select questions; online or onsite proctored 7No prerequisites; 3 or more years in industry, 1 or more on Google Cloud, recommended 7$200 plus tax 7Renewed by exam; Google's renewal table lists a 2-year extension for professional certifications 15
NVIDIA-Certified Associate: Generative AI LLMs 81 hour, 50 to 60 multiple-choice questions; core ML and AI 30%, software development 24%, experimentation 22%, data analysis 14%, trustworthy AI 10% 8A basic understanding of generative AI and LLMs 8$125 82 years; retake the exam to recertify 8
Certified Generative AI Engineer Associate, Databricks 945 scored multiple-choice questions, 90 minutes; online or test center; application development is 30% 9No prerequisites; 6 or more months of hands-on work recommended 9$200 92 years; retake the current exam 9
Artificial Intelligence Engineer (AiE), ARTiBA 16Remote-proctored online exam; six knowledge areas, led by foundational AI/ML theory (22%) and natural language and multimodal systems, including LLM engineering and RAG (20%) 16,17Bachelor's plus 3 years of programming or software engineering; master's plus 2 years; or an experience-based track 16$750, all inclusive 163 years; renewal leads to Senior AiE 16
Predictive AI Engineer, Arcitura 18Online proctored, 110 minutes 18None for the complete exam 18Not on the exam pageNot stated
AI Engineering Professional Certificate, IBM on Coursera 19Self-paced courses and projects (4 months at 10 hours a week); a shareable certificate; no proctored exam described 19Intermediate level 19Listed as included with Coursera Plus; check Coursera for current terms 19Not stated

* Our top pick. We have a business connection with Torchlane.

  • Versions are moving. AWS's updated ML Engineer exam adds generative AI, agentic AI and responsible AI, existing certifications stay valid, and MLA-C01 in English ended on 28 September 2026 5. Google updated its exam for the move from Vertex AI to Gemini Enterprise Agent Platform and says the exam does not directly assess coding skill 7.
  • Two issuers publish less. In what we read, ARTiBA gives eligibility and the fee but no question count or exam length 16,17; Arcitura's exam page gives duration but no fee or validity 18. Ask before you pay.

Is the Azure AI Engineer Associate certification still available?

No. Microsoft's page carries the notice "This certification and the renewal assessment are retired" 10, so it is no longer offered. A current credential Microsoft tags to the same AI Engineer role is the Azure AI Apps and Agents Developer Associate (AI-103), which expects Python experience and builds on Microsoft Foundry 4. Check the exam code before buying a course: AI-102 is the retired one.

Do employers hiring AI engineers ask for these certifications?

Almost never, in the postings we read. Of 119 AI Engineer postings from 31 employers, one lists certifications as a qualification, and only as a nice-to-have: Cloudflare's Senior Oracle Security and AI Engineer role, whose example is Oracle Cloud Infrastructure and Security certifications, not an AI credential 11. Five more, three at Databricks and two at Elastic, use the word only in the pay-range paragraph, as one of several factors that can affect pay (Databricks example: 20). None of the other 118 asks candidates to hold a certification, and none names AI-103, an AWS, Google Cloud, NVIDIA or Databricks exam, or ARTiBA. OpenAI and SpaceX supply 54 of the 119; counted by employer, the result is the same.

Table 2 scrolls sideways inside its frame.

Six AI Engineer postings, read on 23 October 2026.
Employer and titleExperience the posting asks forCertification mentioned
Stripe, AI Engineer 214 or more years shipping production systems; building and deploying AI agents or LLM-powered toolsNone
Brex, AI Engineer, Workbench 225 or more years delivering customer-facing applications; production agent experienceNone
Anthropic, Applied AI Engineer, Startups 234 or more years as a software engineer, forward deployed engineer or technical founder; production LLM applicationsNone
Databricks, Applied AI Engineer 242 to 8 years of engineering experienceNone as a qualification
Datadog, Staff AI Engineer 25A computer science or related degree or equivalent experience; shipped LLM-powered or GenAI applicationsNone
Cloudflare, Senior Oracle Security and AI Engineer 118 or more years in the Oracle ecosystemNice-to-have: relevant professional product certifications, such as Oracle's

What they ask for is experience: years of production engineering and shipped LLM work. Widen the title and an exception appears: Samsara's AI Engineering Manager posting lists AWS Solutions Architect Professional or DevOps Engineer Professional among the things an ideal candidate also has 26. Our sample is mostly technology employers, so it may understate how often other kinds of employer, consultancies and cloud partners among them, ask for exams.

Older research agrees. Georgetown's Center for Security and Emerging Technology analyzed US AI job postings from 2010 to 2020 and found little employer demand for AI certifications; many large technology companies that market their own were largely not using them to qualify applicants for AI jobs 12. That data ends in 2020 and says nothing about postings today, hence our fresh check.

Partners are one place where certified headcount is formal. AWS's services partner tiers list certified-individual minimums: Select, 2 foundational and 2 technical; Premier, 10 and 25, with at least 10 of the technical ones at professional or specialty level 13. At a consultancy or AWS partner, an employer may ask for, or pay for, an exam.

Which AI engineer certification fits which situation?

  • You already build on one cloud. Take that cloud's exam: AI-103 for Azure, the Machine Learning Engineer Associate for AWS (the Generative AI Developer Professional once you have the 2 or more years AWS lists), the Professional Machine Learning Engineer for Google Cloud, or the Generative AI Engineer Associate on Databricks 4,5,6,7,9.
  • You want a first, lower-barrier exam. NVIDIA's associate lists only a basic understanding of generative AI and LLMs 8. Databricks lists no prerequisites but recommends six months of hands-on work 9. AWS's AI Practitioner is foundational, $100 and 90 minutes 27.
  • You work for a consultancy or cloud partner. Ask which exam counts toward your employer's partner status, and whether the employer pays 13.
  • You want a credential not tied to one cloud. ARTiBA's AiE has published eligibility rules and a $750 fee, and ARTiBA says it does not develop, market or promote any AI platform or tool 16. We found no AI Engineer posting naming it, so ask target employers first.
  • You are a working professional moving toward AI work, growing into a bigger role through it, or making your current expertise more valuable with it. Our top pick: Torchlane's AI FDE training builds applied skill through 3 to 6 client-style projects and a capstone, with project work assessed against seven published categories, and no coding experience is needed to apply 1,2,3. If you come from business or operations, it has you build and evaluate AI-assisted workflows around the expertise you already have; if you come from engineering, it adds customer discovery, client communication and a business-focused portfolio 1. It also includes module quizzes and a timed final exam, and it issues its own nonaccredited academy credentials rather than a cloud vendor certification 3. Its AI FDE page lists two exam vouchers, and its training page says the optional AWS and Google Cloud exams are separate vendor certifications with their own requirements 1,3. Torchlane calls the program career preparation, says neither its assessment nor the vendor exams should be presented as accreditation or a guarantee of employment, and says completing a program does not guarantee hiring, a salary or a promotion 1,2. It will not replace a vendor certification a posting names.

We have a business connection with Torchlane.

Certification, certificate or portfolio: what does each prove?

In its Current Population Survey, the US Bureau of Labor Statistics defines a certification as a credential from a non-governmental body, earned when an examination shows the designated knowledge, skills and abilities, and notes that such credentials are time-limited and renewed periodically 28. The vendor exams above fit that definition. The IBM program is a course credential: finish courses and projects, earn a shareable certificate 19.

They prove different things. An exam shows you can answer questions about one platform's services; a course certificate shows you finished a course. Neither shows you can scope a problem and ship a system, which is what the postings above describe 21,23,22. Deployed work you can walk an interviewer through is evidence of that kind. For a ranking by goal, see our guide to the best AI certifications; for whether the fee pays back, read is an AI certification worth it.

What should you check before you pay for an exam?

  1. Which platform do your target postings name? Read five postings for the role you want and count the clouds and tools. Buy the exam for the one that appears most.
  2. What experience does the exam assume? Google recommends 3 or more years, AWS lists 2 or more for its Professional exam, Databricks 6 months 7,6,9. Sitting an exam far above your experience wastes the fee.
  3. What does it cost to keep? Microsoft renewals are free 14; NVIDIA and Databricks credentials last two years and are renewed by retaking the exam 8,9; AWS's ML Engineer credential lasts three years 5.
  4. Is the exam about to change? Microsoft retired one, AWS has an updated exam in beta, Google updated its exam 10,5,7. Study from the current exam guide.

FAQ: questions readers ask

Do you need a certification to become an AI engineer?

Usually not, judging by the postings we read. In 119 AI Engineer postings from 31 employers that we read on 23 October 2026, none required a certification, and the postings we read in detail asked for production engineering experience and shipped LLM work instead. Georgetown's CSET reached a similar conclusion for AI job postings from 2010 to 2020. An exam may still be worth it if your employer pays for it, your employer is a cloud partner, or you want structured study.

Is the Azure AI Engineer Associate (AI-102) still available?

No. Microsoft's page says the certification and its renewal assessment are retired. A current Azure credential tagged to the AI Engineer role is the Azure AI Apps and Agents Developer Associate (AI-103), a 120-minute proctored exam that expects Python experience and builds on Microsoft Foundry.

How much does an AI engineer certification cost?

On 23 October 2026: NVIDIA's generative AI associate exam $125, AWS Machine Learning Engineer Associate $75 beta pricing for the updated exam and $150 for the previous one (no longer offered in English), Google Cloud Professional Machine Learning Engineer $200, Databricks Generative AI Engineer Associate $200, AWS Generative AI Developer Professional $300 and ARTiBA's AiE $750. Microsoft sets AI-103 pricing by country or region. Check the issuer, because fees change.

How long do AI engineer certifications last?

It varies. Microsoft associate certifications renew yearly through a free online assessment. NVIDIA and Databricks credentials are valid for two years and are renewed by retaking the exam. The AWS Machine Learning Engineer Associate and ARTiBA's AiE are valid for three years. Google Cloud's professional certifications, including the Professional Machine Learning Engineer, are renewed by exam, which Google's renewal table lists as adding two years.

Can I earn an AI engineer certification without coding experience?

The engineer-level vendor exams assume some: Microsoft's AI-103 expects experience developing apps in Python, and Google recommends three or more years of industry experience while noting that minimum Python and SQL proficiency lets you read the code in its questions. AWS's foundational AI Practitioner exam is aimed at a wider audience, including line-of-business roles. Torchlane's AI FDE training says you can apply without coding experience and that you will learn to work with code; its credentials come from Torchlane's own academy, are nonaccredited and rest on module quizzes, a timed final exam and assessed projects rather than a vendor exam, and Torchlane says it is not a substitute for the engineering experience senior roles require. We have a business connection with Torchlane.

Sources

  1. Torchlane. “AI FDE training: forward deployed AI engineering.” Accessed 23 Oct 2026. Back to text
  2. Torchlane. “AI skills assessment and credentials.” Accessed 23 Oct 2026. Back to text
  3. Torchlane. “FDE training and AI certification.” Accessed 23 Oct 2026. Back to text
  4. Microsoft Learn. “Microsoft Certified: Azure AI Apps and Agents Developer Associate.” Published 2026-08. Accessed 23 Oct 2026. Back to text
  5. Amazon Web Services. “AWS Certified Machine Learning Engineer - Associate.” Accessed 23 Oct 2026. Back to text
  6. Amazon Web Services. “AWS Certified Generative AI Developer - Professional.” Accessed 23 Oct 2026. Back to text
  7. Google Cloud. “Professional ML Engineer Certification.” Accessed 23 Oct 2026. Back to text
  8. NVIDIA. “NVIDIA-Certified Associate Generative AI LLMs Certification Exam.” Accessed 23 Oct 2026. Back to text
  9. Databricks. “Databricks Certified Generative AI Engineer Associate.” Accessed 23 Oct 2026. Back to text
  10. Microsoft Learn. “Microsoft Certified: Azure AI Engineer Associate.” Published 2026-06. Accessed 23 Oct 2026. Back to text
  11. Cloudflare. “Senior Oracle Security and AI Engineer (job posting).” Accessed 23 Oct 2026. Back to text
  12. Center for Security and Emerging Technology, Georgetown University. “U.S. Demand for AI Certifications: Promise or Hype?.” Published 2021-06. Accessed 23 Oct 2026. Back to text
  13. Amazon Web Services. “AWS Partner Services Tiers.” Accessed 23 Oct 2026. Back to text
  14. Microsoft Learn. “Microsoft Certification Renewal.” Published 2025-01. Accessed 23 Oct 2026. Back to text
  15. Google Cloud Certification Help. “Certification Renewal Options.” Accessed 23 Oct 2026. Back to text
  16. ARTiBA. “Artificial Intelligence Engineer (AiE) Certification.” Accessed 23 Oct 2026. Back to text
  17. ARTiBA. “ARTiBA Certification Examination.” Accessed 23 Oct 2026. Back to text
  18. Arcitura. “Predictive AI Engineer Certification.” Accessed 23 Oct 2026. Back to text
  19. Coursera. “IBM AI Engineering Professional Certificate.” Accessed 23 Oct 2026. Back to text
  20. Databricks. “AI Engineer - Forward Deployed Engineering (AI FDE) (job posting).” Accessed 23 Oct 2026. Back to text
  21. Stripe. “AI Engineer (job posting).” Accessed 23 Oct 2026. Back to text
  22. Brex. “AI Engineer, Workbench (job posting).” Accessed 23 Oct 2026. Back to text
  23. Anthropic. “Applied AI Engineer, Startups (job posting).” Accessed 23 Oct 2026. Back to text
  24. Databricks. “Applied AI Engineer (job posting).” Accessed 23 Oct 2026. Back to text
  25. Datadog. “Staff AI Engineer - Agent Observability (job posting).” Accessed 23 Oct 2026. Back to text
  26. Samsara. “AI Engineering Manager (job posting).” Accessed 23 Oct 2026. Back to text
  27. Amazon Web Services. “AWS Certified AI Practitioner.” Accessed 23 Oct 2026. Back to text
  28. U.S. Bureau of Labor Statistics. “FAQs about data on certifications and licenses (CPS).” Accessed 23 Oct 2026. Back to text

How this was researched. Written by the AI Certification Reviews editorial team from the sources listed above, each checked on the date shown. Where our facts come from.

More from AI Certification Reviews