How I Earned the AWS Gold Jacket

I passed AWS Certified Generative AI Developer – Professional on January 2nd, which finished the set. The jacket itself took until May to turn up.

Technically it was already finished. Advanced Networking – Specialty in December 2024 was the last one on the board at the time, and for about a year after that I held everything AWS offered. Then they released a new exam and I didn’t anymore.

The first one was Solutions Architect – Associate in May 2014. The certification program was only about a year old then and there were a handful of exams total, so “all of them” meant something a lot smaller than it does now.

Eleven and a half years, start to finish.

Not eleven and a half years of studying, though. For most of that stretch I wasn’t taking exams at all.

The list

• Solutions Architect – Associate May 2014
• Developer – Associate Oct 2014
• SysOps Administrator – Associate Apr 2015
• DevOps Engineer – Professional Aug 2015
• Solutions Architect – Professional May 2016
• Big Data – Specialty Dec 2016
• Security – Specialty Jan 2018
• Machine Learning – Specialty May 2019
• Data Analytics – Specialty Jan 2020
• Cloud Practitioner Feb 2020
• Alexa Skill Builder – Specialty Nov 2020
• Database – Specialty Nov 2020
• Data Engineer – Associate Dec 2023
• AI Practitioner Sep 2024
• Machine Learning Engineer – Associate Oct 2024
• Advanced Networking – Specialty Dec 2024
• Generative AI Developer – Professional Jan 2026

And the same thing with dates attached.

I sat Security – Specialty twice, so the real figure is 1,200 questions. Somewhere north of forty hours in a testing chair, spread across eleven and a half years.

Those counts use today’s formats, where Foundational, Associate, and Specialty exams are 65 questions and Professional exams are 75. The 2014 and 2015 versions were shorter, so the number is a little generous, and I’m not going back to work out the exact one.

Three of those were betas, sat before the exam was generally available. AI Practitioner, Machine Learning Engineer, and Generative AI Developer. Beta sittings are discounted, the questions are rougher, and the score takes months to show up.

Three years of nothing

November 2020 to December 2023. Not a single new AWS exam.

By the end of 2020 I’d been through most of what AWS offered. Advanced Networking was still sitting there and I kept not getting around to it. So I went sideways instead. Eight Google Cloud certifications landed in roughly that window, plus Terraform, CKAD, and Security+.

The chart goes flat, and I was arguably learning more during the flat part than during the climbs on either side of it. Certifications track what exams exist, not what you know.

Then AWS started shipping again. Data Engineer – Associate in December 2023. AI Practitioner and Machine Learning Engineer three weeks apart in the fall of 2024. Advanced Networking that December, finally.

Retired exams

Three of the seventeen are for exams that don’t exist anymore.

Big Data – Specialty was replaced by Data Analytics – Specialty, which I passed in January 2020, and which AWS then retired as well. Alexa Skill Builder is gone too. So two of the things on that list are things nobody can get certified in now, and one of them I only took because its predecessor had been discontinued.

Which makes “all of them” a slightly strange target. The set moves underneath you.

Recertification

AWS certifications last three years. Seventeen certifications means seventeen expiration dates, which sounds worse than it is.

One rule makes it manageable. Passing a Professional level exam recertifies your Associate and Foundational certifications, and passing an Associate recertifies the Foundational ones.

You can see it happen in my transcript. The Generative AI Developer pass on January 2nd pushed Cloud Practitioner, Data Engineer – Associate, Machine Learning Engineer – Associate, and AI Practitioner all out to January 2029.

Specialties don’t work that way, so those you sit again. I retook Security – Specialty two days after the original expired, which I would not hold up as a scheduling strategy.

Plan around your Professional exams rather than around your renewal dates.

Writing the exams instead of taking them

Somewhere in the middle of all this I ended up on the other side. I did subject matter expert work with AWS Certification starting in 2021, and lead SME work on an Associate level exam in 2022.

Exam development is less glamorous than it sounds. A lot of it is arguing about whether a wrong answer is wrong for the right reasons, and whether a question measures the skill it’s supposed to or just measures careful reading.

It made me better at taking them, which was probably inevitable. It also made me more skeptical about what they prove. A good exam is deliberately narrow. None of them are trying to tell you whether someone can run a migration.

Skip the courses. Open the console.

The question I get more than any other is which course to buy.

Don’t. Save the money.

Hands on practice is the only thing I’ve seen reliably work. Video series, practice exam bundles, bootcamps, none of it produces someone who can do the job. Some of it will get you through a test, which is a different accomplishment.

Open your own AWS account and build something. Not a tutorial. Following tutorials teaches you to follow tutorials. Pick something you actually want to exist and make it run, and you’ll end up learning in all the places a tutorial would have smoothed over. IAM denies you and you have to work out why. A NAT gateway bills you quietly for a month. Something works in us-east-1 and doesn’t in eu-west-1.

That knowledge stays. I couldn’t tell you much now about the material I studied for Big Data in 2016. I remember every mistake I’ve made in a live account.

Set a budget alarm before you start, stay in the free tier where you can, and tear down what you spin up. The bill is part of the curriculum.

For the exam itself, the AWS documentation and the service FAQs beat anything you can pay for and they cost nothing. Read the exam guide, find the parts you can’t confidently explain, go learn those. Studying is for filling gaps rather than for building the foundation.

I’m blunt about this because it shows up in the work. Someone who got there on course material alone has the badge and not much behind it, and on a real engagement that’s obvious inside a day.

If you’re starting now

• Skip Cloud Practitioner if you already work in this field. I took it sixth, in February 2020, six years after Solutions Architect – Professional, which should tell you how necessary it was. If you’re new to AWS it’s a good on-ramp.
• Take the three Associates close together. Solutions Architect, Developer, and SysOps overlap enough that doing them in one stretch means learning the material once.
• Sit the betas. Cheaper, and you get to be early on things like AI Practitioner.
• Watch the renewal rule rather than the renewal dates.
• Assume the list will change on you. Three of mine are for exams that no longer exist.

Was it worth it

Yes, though not for the reason people assume.

The certifications have never won me an engagement. Nobody has hired me because I hold seventeen of anything. What the breadth buys is that projects stall less often. A relevance problem turns into a VPC problem, or an IAM problem, or a conversation about instance families and what they cost, and I can just have that conversation instead of going to find someone who can.

That’s the return. The jacket is a bonus, and it is extremely shiny.

If you’re working on something on AWS and want another set of eyes on it, please reach out.