The AWS certification catalog now has 12 active credentials, and Amazon added two new specialty exams in late 2025. If you’re sitting at your desk wondering which one to actually study for, the answer almost certainly isn’t “all of them”, but the internet will try to sell you that story anyway.
I’ve talked to engineers at a range of companies, from scrappy Series A startups to large consultancies, and the pattern I keep hearing is roughly the same. People spend 4 months studying for the wrong cert, land a job that doesn’t care about it, and regret not getting the Solutions Architect Associate first. That’s not universal, but it’s common enough to mention up front.
The cert tier structure, honestly explained
AWS organizes certifications into three tiers: Foundational, Associate, and Professional, plus a separate Specialty track. The Foundational level has one cert (Cloud Practitioner). The Associate level has three: Solutions Architect, Developer, and SysOps Administrator. Professional has two: Solutions Architect Pro and DevOps Engineer Pro. Specialties cover areas like Machine Learning, Security, Networking, Database, and Data Analytics.
Most hiring managers in the US don’t require certs, but they do notice them at the Associate level and above. The Bureau of Labor Statistics projects cloud-related roles growing 15% through 2032, faster than most tech categories, which is part of why AWS cert pass rates and sitting volumes keep climbing.
Cloud Practitioner is useful if you need to demonstrate baseline AWS literacy to a non-technical interviewer or if you’re completely new. If you already write any code or manage any infrastructure, skip it. It’s a 90-minute multiple-choice exam and the knowledge doesn’t transfer that well to the Associate exams anyway.
Which Associate cert to get first
Solutions Architect Associate (SAA-C03) is the one most engineers should target first. It covers the broadest slice of AWS services, it’s what recruiters search for on LinkedIn, and it gives you enough architectural vocabulary to make DevOps and Developer Associate material easier afterward.
Developer Associate (DVA-C02) makes sense if you’re already in a developer role and your company is heavily invested in services like Lambda, API Gateway, and DynamoDB. The exam leans more toward SDK usage and CI/CD pipelines than the SAA does.
SysOps Administrator Associate is, frankly, the hardest of the three. It involves a lab component where you actually have to perform tasks in a live AWS console. Worth getting if you’re in an ops or platform engineering role, but don’t start here.
The Stack Overflow 2024 Developer Survey found AWS was the most used cloud platform at 48% of respondents, which helps explain why SAA specifically shows up on more job posts than Azure or GCP equivalents.
How long does studying actually take
For Solutions Architect Associate, most people with some cloud exposure need 6 to 10 weeks studying roughly an hour a day. Adrian Cantrill’s course (cantrill.io) is the best structured resource I’ve seen. Stephane Maarek’s Udemy course is cheaper and covers the same material, though with less depth on edge cases.
Practice exams matter a lot. The real exam has 65 questions in 130 minutes, and some questions are long scenario-based problems. Tutorials Dojo (Jon Bonso’s practice exams) are closer to the actual difficulty than the official AWS sample questions.
I don’t know exactly what the current first-attempt pass rate is, AWS doesn’t publish it. Anecdotally, people who do 3 or more full practice exams before sitting pass at a much higher rate than those who don’t. That’s not a revelation, but it bears saying.
The Professional tier is a different animal
Solutions Architect Professional requires at least a year of hands-on AWS experience. This isn’t a suggestion, the exam tests for decision-making at a level that’s basically impossible to fake without having actually designed and debugged systems. The questions involve 40-word scenarios with 5 answer choices where two are plausibly right.
If you’re aiming for SAP-C02, give yourself 3 to 5 months after passing the associate, not before. People who rush it tend to fail twice, which wastes both money (the exam costs $300) and time.
Specialty certs and when they’re worth it
The Machine Learning Specialty (MLS-C01) has seen a surge of interest since 2023, and it shows in the study material quality, there’s now a decent ecosystem of courses and practice tests. If you’re working in a data or ML engineering role, it signals real depth.
Security Specialty is valuable if your company is SOC 2 or FedRAMP focused. Networking Specialty is niche but commands a salary premium at cloud consultancies. Database Specialty is honestly a bit underappreciated given how much time senior engineers spend troubleshooting RDS and DynamoDB issues.
The newer AI Practitioner cert (released late 2024) is lower stakes, think of it as a Cloud Practitioner equivalent for generative AI topics. It’s not a substitute for the ML Specialty if you want to signal genuine ML engineering depth.
One thing Craqly is useful for here
If you’re actively interviewing for cloud roles alongside studying, Craqly’s interview copilot can help you work through live AWS architecture questions during technical interviews. It’s not a study replacement, but for actual job interviews where you’re asked to walk through a system design on AWS, having a tool that can pull up relevant service comparisons in real time matters.
The order I’d actually recommend
Start with SAA-C03. Then either DVA-C02 or Security Specialty depending on your role. Then SAP-C02 if you want the professional credential. That path takes most engineers about 18 months at a reasonable pace, leaves you with three meaningful credentials, and maps to what employers actually ask about.
Whether the ML Specialty belongs in that path depends on whether your work touches ML pipelines at all. If it doesn’t, it’s probably not the best use of 3 months of evenings.