MLA-C02 career: ML engineer roles on AWS
MLA-C02 maps to a short list of titles that AWS itself publishes: ML engineer, MLOps engineer, LLMOps engineer, data engineer, backend software developer and data scientist. The “LLMOps engineer” tag is new compared with the previous version of the exam, and it signals what AWS now expects from the role: running foundation models in production, not only classic models. The wording comes from AWS’s certification page, and the full task list is in the MLA-C02 exam guide.
From “backend developer” to “LLMOps engineer”: the six titles on AWS’s list
The tags describe where people start, not where the credential ends. Mapped to the work in the exam guide:
| AWS tag | What the exam gives you evidence for | Likely gap to close |
|---|---|---|
| ML engineer | The full chain: data preparation, training, deployment, monitoring | Depth in one modelling area |
| MLOps engineer | CI/CD for models, orchestration, drift monitoring, endpoint scaling | Team-level platform design |
| LLMOps engineer | Bedrock model selection, RAG pipelines, agent deployment, token cost control | Evaluation practice beyond the exam |
| Data engineer | Ingestion, validation, feature preparation (the 28% domain) | Modelling and serving |
| Backend software developer | Calling and hosting models, securing endpoints | Data quality and bias work |
| Data scientist | Training and evaluation inside a production setting | Pipelines, IaC, cost management |
AWS also states the experience bar behind these tags: at least one year with Amazon SageMaker AI, Amazon Bedrock and related services, and one year in a role such as backend developer, DevOps developer, data engineer or data scientist. This guide’s prerequisites page turns that into a checklist, so it is not repeated here.
What the exam guide says a holder is not expected to do
AWS publishes four job tasks as out of scope for the target candidate:
- designing and architecting full end-to-end AI and ML solutions
- setting up best practices and guiding ML strategies
- handling integration with a wide array of services or new tools and technologies
- working deeply in two or more ML domains
Read that as a limit on what a recruiter should infer. A holder has shown they can build and operate within a design, not that they can own the design, lead an ML platform strategy or specialise in, say, both computer vision and NLP. Roles titled “ML architect”, “staff ML engineer” or “head of MLOps” ask for exactly those things, so the credential supports an application for them but does not evidence them. A repository, a deployed pipeline or a written design is stronger evidence for such roles.
What changed from MLA-C01 to MLA-C02 for people changing jobs
The domain weights barely moved: data preparation stays at 28%, model development drops from 26% to 24%, deployment rises from 22% to 24%, and monitoring and security stay at 24%. What changed is content. AWS’s comparison page lists additions such as vector databases, embedding models, RAG document preparation, Amazon Bedrock knowledge bases, agent deployment pipelines, agent state management, prompt management, LLM-as-a-judge evaluation and Bedrock Guardrails. It lists removals such as SageMaker Neo edge optimisation and bring-your-own-container.
For a job seeker this matters in two ways. A holder of the older MLA-C01 has a credential that “remains active through its original expiration date”, but it does not show the generative AI tasks above. And a candidate who learned edge deployment may find that skill no longer examined. The model development and deployment pages cover how the new content is tested.
Next steps: check AWS’s own recommendation before following it
AWS’s certification page recommends AWS Certified Machine Learning - Specialty for “professionals looking to dive deeper into machine learning”. That advice is stale: the Specialty page says the certification is being retired and that the last day to take it was 31 March 2026. Holders keep an active credential for three years from the date earned, but no new candidates can sit it. This guide reports the discrepancy because a reader following the MLA-C02 page alone would plan around an exam that no longer exists.
Realistic next steps that do exist:
- AWS Certified Generative AI Developer - Professional. AWS lists 300 USD, 180 minutes and 75 questions, recommends two or more years building production-grade applications plus one year of hands-on generative AI work, and names MLA-C02 as a helpful earlier credential. It is also a renewal route, covered on the renewal page.
- Data Engineer or Solutions Architect Associate on AWS. Sensible if the out-of-scope list above describes your gap: breadth and design rather than more ML.
- AIF-C01 is not a step up. AIF-C01 is for people who use AI without building it, so it is rarely a step after MLA-C02. The comparison explains why.
- Cross-vendor equivalents. Teams that run both clouds look at AI-300 for Azure machine learning operations or the Databricks ML Associate for the lakehouse side.
The AWS AI certification path shows how these fit together.
Where to look up pay and demand
This page quotes no salary or hiring numbers. For neutral occupation descriptions, search the US Bureau of Labor Statistics Occupational Outlook Handbook for computer and information research scientists, data scientists and software developers, and compare with postings in your own market. Whether the cost of the exam is justified is covered in is MLA-C02 worth it, and the certification costs page lists fees across vendors.