Review of the AWS Microcredentials

Apart from certifications, AWS now offers so-called microcredentials that one can achieve to verify one’s skill. Contrary to theoretical multiple-choice questions, microcredentials can be earned by passing lab-based practical exams in which you have to complete a series of tasks in an unproctored but timed setting. The intention is that these serve as complementary credentials to assess one’s ability to practically perform tasks in AWS. After all, theory is supposed to be applied in practice, right?

In this article, I will describe my experience as I passed all four current microcredentials and will share my opinion about their use.

After completion, you will get Credly badges, and also have them listed on your Skills profile here where the credentials can be verified.

History

The theoretical certifications from AWS are challenging, but often get criticized for only testing theory. Admittedly, many of these can be achieved purely by studying hard without actually building something. Employers and senior engineers alike will however wonder if you are actually capable of building infrastructure, or to perform troubleshooting tasks under high pressure. So is there any way to prove that you are actually capable of doing these hands-on tasks?

A few years back, the AWS Certified SysOps Administrator certification exam contained a practical component where you had to set up actual infrastructure as part of the exam. But eventually this practical component was removed and didn’t return. I don’t know exactly why, but I also remember that this part was pretty flaky. Anyway, instead, AWS came up with these so-called microcredentials: Dedicated lab-based exercises where you can verify your hands-on skills.

These microcredentials used to be locked away in a SkillBuilder subscription from AWS’s learning platform that was mandatory to have before you could attempt these. As of recent however, these microcredentials are now all free for anyone to attempt.

Topics

To get started, you have to create a SkillBuilder account and access the microcredentials on this page. You can then get started right away. There are four microcredentials available and a Preview Experience exercise, that serves as a trial lab so you can get used to the interface.

• Serverless: Setting up and troubleshooting serverless infrastructure with the main focus on Lambda, API Gateway and DynamoDB.
• Application Networking: Configure networking functionality with VPC Lattice, ECS, API Gateway, CloudFront, and monitor with CloudWatch Alarms.
• Agentic AI: Orchestrate a multi-agent AI infrastructure with Bedrock.
• Incident Response: Respond urgently to an attacker abusing your EC2 servers for malicious purposes. Gather evidence using Athena, secure your environment with Security Groups, and use IAM to banish the attacker.

Prerequisites & Difficulty

The labs have no official prerequisites: You can get started with these any time, for free. All you need is your computer and 90 to 100 minutes of your free time (differs per lab). If you fail any microcredential, then you can retake them after 25 days. Note that the exercises will be different every time.

Compared to the certifications, I would say that these labs are pretty much on the Associate level in terms of difficulty. The instructions are straightforward and you are free to use any public resource on the internet (including AI, more on this later) to help you. The most difficult part is probably managing time, because you have to complete 6–7 exercises and you simply don’t have the time to look up everything. So some prior knowledge is necessary for you to perform the tasks and navigate around the console.

If you are a senior engineer with multiple years of experience and/or hold one or more professional certifications, then you should be able to pass all of these microcredentials without any prior preparation.

If you are a junior engineer, then I suggest to get at least the Solutions Architect Associate certification so you will have a minimal base understanding of the purpose of each of the services that you will encounter. Also you can use the SkillBuilder platform to prepare accordingly, using SimuLearn and CloudQuests.

Getting Started

As mentioned, each of the labs is unproctored and takes 90 to 100 minutes each, that you can take anytime. During these labs, you have to complete 6 or 7 exercises that are mostly also dependent on each other so you often have to complete the prior one before you can move on to the next.

I will also stress that your time management is critical. Even if things go well, you may eventually end with 5 or 10 minutes left on the clock. This means that every minute counts. So although you can take a break anytime if you want to, I strongly suggest going to the bathroom and making yourself a drink before getting started so you can remain seated during the entire lab session. Put your phone in flight mode, and if you want to listen to music, then turn on a radio where you don’t have to manually select the next song every time.

Before starting, you are asked to agree to keep the lab content confidential and not to record anything. It is however explicitly allowed (and even encouraged) to use public resources and an AI to support you to solve the tasks. After all, in the real world you would do that too.

Don’t get too excited though: While you can ask an AI for help, you are not allowed to copy over the entire exercise. Also the AI can only instruct you what to do, and not do the actual work for you. To avoid being given wrong instructions, you will need good prompting skills. Cleaning up wrongly configured infrastructure costs time and you might even accidentally delete a critical resource that you need later. Furthermore, you will not have the time to prompt your way through everything. There simply won’t be enough time.

My Experience

I will share my experience with each of the labs, without giving too much away about the actual content. This is purely subjective.

General

As mentioned, you will have 6 or 7 exercises to complete, and some, if not all, will be dependent on each other. Each microcredential has roughly the same difficult level. Not all exercises are equally difficult or time consuming, but you will have to manage your time. You should complete the exercises in order. If you do get stuck however, you may want to move on.

The first exercise is usually the hardest and the most time-consuming because you have to get used to the environment, understand the exercise, and do some initial configurations. Don’t worry about this: Some of the follow-up exercises will cost less time so you can afford spending more time on the first exercise.

The exercises don’t have trick questions. With each exercise you have to complete several tasks, which are mostly straightforward to do, but you can still slip up. Overall it is well doable. Sometimes it is better to look at the success criteria (if provided) than the actual task description to understand what you actually need to do.

Each exercise has a checkpoint: You can have your exercise checked to verify that your solution is complete. If it is incomplete, the exercise will tell you where something is still wrong (in red letters). If it is complete, you will receive a positive feedback (in green letters) and you can move on to the next exercise.

Some exercises will encourage you to try some end-to-end testing yourself. Mostly you do not actually have to do this, and I would even suggest not to do this, because there is simply not enough time. If you think you are done, check the exercise, and if it is marked as complete, simply move on to the next exercise without looking back.

Once all of your exercises are complete, you can end the exam and you will get your badge nearly instantly. I’m not sure if all exercises have to be completed successfully in order to pass the exam, I think not, but I haven’t seen any cutoff of e.g. having to complete at least 80% to pass. In any case, I suggest that you just try to complete all exercises.

Serverless Lab

As the name suggests, you will be working on a serverless infrastructure. You may have to set up a few new things, or perform some troubleshooting to get a pipeline of services to work. Expect it to be heavy around Lambda, API Gateway, and DynamoDB, among others.

I think this lab is actually slightly easier than others but you can still slip up. For example, if you have to update a Lambda version, then don’t forget to deploy the new API version in API Gateway to the right stage with the right configuration.

Perhaps some reassurance: You do not have to do any programming. You should however be able to read some Python. Most Lambda-related exercises will be around setting up configurations (permissions, environment variables) and connecting it to other services.

Application Networking Lab

Here the focus is on content delivery: Expect to do some work around ECS and configuring CloudFront, a VPC Lattice, and Load Balancers. You may also have to configure CloudWatch Alarms to monitor your setup. Essentially you’ll be deploying a full-stack application and making it available.

Possible caveats: ECS and Fargate can feel a little confusing at first. It’s easy to apply the wrong configuration. The rest should be straightforward.

Agentic AI Lab

Now you’ll actually get to work on a full-fledged agentic GenAI infrastructure. This is mainly centered around Bedrock and configuring a multi-agent infrastructure properly, and integrate it with a broader infrastructure. Things like Guardrails and Knowledge Bases can be part of the mix.

Also here the beginning is the toughest part because you might not exactly know where to get started. Let the AI guide you here (more on this below). Overall I think this is the most difficult microcredential of all, even though the difference in difficulty to the other labs is minimal.

Incident Response Lab

A classical scenario: A hacker has entered your AWS environment, stolen some data, and is now abusing your computing capacity for malicious purposes. Your task is to do forensical work by securing evidence and ensuring the hacker no longer has access to your environment. Services in the mix include S3, EC2, Security Groups, IAM and Athena.

Get Marijn Scholtens’s stories in your inbox

Join Medium for free to get updates from this writer.

Remember me for faster sign in

Athena can be a bit unintuitive if you haven’t used it before. Be sure to use any lab-provided resources to help you here. You can also use your AI to help you to set up custom SQL queries to gather data.

Tips for using AI

Firstly, I suggest using two monitors. You will have three windows: One with the lab exercise, one with the actual lab console, and one with your favorite AI open. Even an anonymous ChatGPT session will suffice for these labs. You will probably not need to use Google or the AWS documentation because an AI is sufficient and searching through documentation yourself would probably take too much time.

Firstly, set the right context. You can start with a prompt like this:

I want to take the AWS microcredential assessment for (..) which is unproctored and where using AI for supportive purposes is explicitly allowed. Can you support me if I have any questions? Be concise and to-the-point in your answers, and also I will not have to make any IAM modifications unless explicitly stated.

The conciseness is important so you don’t have to weed through long texts to get what you need. Then during the exam you can ask questions like:

How do I set up a multi-agent configuration in Bedrock? Walk me through the steps

I have updated my Lambda function, how do I publish the new version?

I am having trouble with (..) what are some things I can do first for troubleshooting and finding the problem?

Do not bother trying to ask the AI how a service works and what it does. Mainly use it to walk you through the steps to make specific configurations, like where to click and what to select.

Remember: Do not copy entire exercise texts as this violates the lab code of conduct. (Copying actually isn’t even possible — texts from the lab are not copyable so you’d have to type them over, which costs time).

Bugs and Problems

I did not encounter any technical difficulties. The lab environment seemed pretty stable and I was able to complete all tasks in all exercises.

In one of the labs however there was a conflict between two exercises. In one exercise I had to delete a resource but in a later exercise I was instructed to recover data from exactly that resource. Fortunately, you can ‘recreate’ this data yourself with a small workaround. I reported this to AWS in the lab feedback form.

Are these Microcredentials Valuable?

The million-dollar question: How valuable are these microcredentials really? Will these get me hired?

While the microcredentials have their merit, I will be honest and say that at the same time, the real value is rather limited.

Completing these microcredentials shows:

• You have at least a basic understanding of important AWS services.
• You feel comfortable navigating around the AWS console.
• You can complete tasks and connect services together to set up a working infrastructure.
• You can perform basic troubleshooting.
• You presumably used AI in an effective way to quickly get what you need to complete tasks.
• You can complete everything under time pressure.

At the same time, these microcredentials have their drawbacks:

• The scenarios at hand are not exactly realistic and differ from the real world in many ways. For instance, in the real world you don’t get instant feedback whether your solution is correct.
• The labs are unproctored and free, so you can repeat them often or even get help to “cheat” your way through them.
• There are no architectural challenges to be solved or design decisions to be made.
• You do not need to know how the services actually work, and understand the patterns and limitations behind them.
• You have to perform all configuration manually (“ClickOps”) whereas in most real-world environments, setting up resources is usually done using IaC with e.g. CloudFormation, CDK, or Terraform.

I think they can be a nice stepping stone for aspiring cloud engineers in junior or mid-level roles to showcase that they are actually capable of doing some hands-on work, in addition to the certifications. So overall, they’re a nice addition, but that’s all there is.

However, given that the microcredentials are now free for everyone, there is a chance that employers hiring for AWS roles may ask all of their applicants to achieve any of the microcredentials as a prior application requirement to filter out any applicants who haven’t done any of them. This could be an interesting pre-screening measure that reduces work for recruiters and hiring managers. This may give you an edge over other applicants when applying for roles.

For senior roles, I don’t see much value because your practical experience and professional certifications will weigh much more. On the other hand, it can be a fun exercise to brush up your skills as you probably don’t need to spend any time preparing for it, contrary to certifications.

Validity

The microcredentials have a validity of 1 year, after which you have to renew them by taking the labs again to keep the credential active. It is up to you whether you want to do that or not.

In my personal opinion: I’m not a fan of expirable credentials, because I don’t see the point in constantly having to prove again that you still know how to do something (as if you have suddenly forgotten everything..). It only costs a lot of time, and the maintenance burden is already very big if you’re an engineer that holds multiple Cloud and DevOps certifications that expire every X years. I hope that these microcredentials just become permanent for everyone at some point. Either way, it is very unlikely that I will be renewing these microcredentials. Once you have a solid job, you may as well shift your focus to other things and for example pursue the professional certifications.

Future Microcredentials?

AWS also announced that they want to introduce more microcredentials. I think this is a good thing, because there are many more topics that could be tested practically. If it were up to me, I would make the following suggestions for future microcredentials:

• Landing Zones: Configure a compliant multi-account structure with AWS Control Tower, AWS Organizations and SCPs.
• FinOps: Find ways to reduce the bill by downsizing computing resources and databases.
• Data Engineering: Build a data processing pipeline using S3, Glue, and Redshift.
• Disaster Recovery: Set up failovers and backup plans with databases and load balancers.
• Kubernetes: Launch an EKS cluster and make sure its containerized workloads can be reached over the internet.
• Database Migration: Transfer data from one database to another, the correct way.
• Event Architecture: Set up an event-driven architecture using Lambda, EventBridge, and SQS, among others.
• Security Hardening: Perform overall security hardening with KMS, Secrets Manager, WAF, NACLs and Security Groups.

Conclusion

In this article we have extensively covered the AWS microcredentials that are now available for free for anyone to try out. These are lab-based exercises that verify your ability to take on hands-on work in the AWS console and they are complementary to the theory-based certifications.

At this point in time, four microcredentials are available to pursue and we have briefly gone over all of them to give an idea of how they work and what there is to expect. They can be challenging, but they are also pretty straightforward.

The value of these microcredentials is realistically little at this point. I would consider them to be useful and a nice addition, but we should not expect them to be actual game changers, especially when it comes to job hunting and hiring. The main benefit may be there for junior and mid-level cloud engineers that can use these to give themselves a slight edge over other applicants that may not have them.

I would still encourage senior engineers to give them a try, as a measurement of your practical skill. But after achieving them once, the focus should be back to higher-level certifications and actual practical work that entails architectural designs, decision-making, and automation over basic implementation.