<?xml version="1.0" encoding="UTF-8"?><rss xmlns:dc="http://purl.org/dc/elements/1.1/" xmlns:content="http://purl.org/rss/1.0/modules/content/" xmlns:atom="http://www.w3.org/2005/Atom" version="2.0" xmlns:cc="http://cyber.law.harvard.edu/rss/creativeCommonsRssModule.html">
    <channel>
        <title><![CDATA[Stories by Dmytro Sirant on Medium]]></title>
        <description><![CDATA[Stories by Dmytro Sirant on Medium]]></description>
        <link>https://medium.com/@sirantd?source=rss-cd0af9ee77ad------2</link>
        <image>
            <url>https://cdn-images-1.medium.com/fit/c/150/150/1*38mm9ArBv_nG75LIwII7WA.jpeg</url>
            <title>Stories by Dmytro Sirant on Medium</title>
            <link>https://medium.com/@sirantd?source=rss-cd0af9ee77ad------2</link>
        </image>
        <generator>Medium</generator>
        <lastBuildDate>Wed, 18 Nov 2026 01:32:19 GMT</lastBuildDate>
        <atom:link href="https://medium.com/@sirantd/feed" rel="self" type="application/rss+xml"/>
        <webMaster><![CDATA[yourfriends@medium.com]]></webMaster>
        <atom:link href="http://medium.superfeedr.com/" rel="hub"/>
        <item>
            <title><![CDATA[It’s AI coding era but naming is still hard]]></title>
            <link>https://medium.com/@sirantd/its-ai-coding-era-but-naming-is-still-hard-595591b76375?source=rss-cd0af9ee77ad------2</link>
            <guid isPermaLink="false">https://medium.com/p/595591b76375</guid>
            <category><![CDATA[ai]]></category>
            <category><![CDATA[software-development]]></category>
            <category><![CDATA[software-engineering]]></category>
            <dc:creator><![CDATA[Dmytro Sirant]]></dc:creator>
            <pubDate>Mon, 13 Apr 2026 04:22:57 GMT</pubDate>
            <atom:updated>2026-04-13T04:22:57.117Z</atom:updated>
            <content:encoded><![CDATA[<figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*qilzyWU0ivRUEeEMEwJM3A.png" /></figure><blockquote>“There are only two hard things in computer science: cache invalidation and naming things.” <br>- Phil Karlton</blockquote><p>It’s been about a year since I started using AI daily.</p><p>In my core role as an SRE/DevOps engineer, AI behaves like a very fast and very obedient assistant. It helps me collect data across services via MCP, generate Terraform modules, work with Helm charts, and automate routine tasks.</p><p>There is no “wow” factor here. Just speed.</p><p>I don’t expect creativity. <br>I don’t want initiative. <br>I want predictable output.</p><p>In these scenarios, I already know what the result should look like. I have examples, constraints, and a clear definition of done. AI just accelerates execution. Control stays on my side.</p><p>Then I switch context.</p><p>I start using AI for software development. And everything changes.</p><p>As an AWS Community Builder, I have access to tools like Kiro and Kiro CLI - agentic AI development environments backed by Anthropic (and other) models. These tools are powerful.</p><p>But they expose a different problem.</p><p>Not in AI. In us.</p><h3>The real bottleneck</h3><p>When working with AI on anything non-trivial, the bottleneck is not code.</p><p>It’s how you describe what you want.</p><p>Not syntax. <br>Not frameworks. <br>Not even architecture.</p><p>Language.</p><p>If your instructions are vague, inconsistent, or overloaded with assumptions - the output will reflect that.</p><p>And unlike humans, AI won’t ask clarifying questions by default. It will generate something <em>that looks correct</em> but is fundamentally wrong.</p><p>This is where things start breaking.</p><h3>Naming breaks everything</h3><p>One example from practice.</p><p>I had to modify an old project that wasn’t maintained for ~8 years. The goal was simple: update dependencies, remove legacy functionality, and add new features.</p><p>The problem was not the code.</p><p>The problem was naming.</p><p>The same concept - a media file - was called:</p><p>- `track`<br>- `item`<br>- `product`</p><p>Depending on the part of the codebase (mobile apps, backend, frontend), they were developed by different freelancers at different times.</p><p>From a human perspective, this is annoying.</p><p>From an AI perspective, this is chaos.</p><p>The model tries to infer relationships. It guesses. Those guesses propagate into generated code, queries, and logic.</p><p>Before doing any actual work, I had to:</p><p>1. Recall the real domain model <br>2. Standardise terminology <br>3. Refactor naming across the codebase</p><p>Only after that AI became useful again.</p><h3>Non-native constraint (and advantage)</h3><p>As a non-native English speaker, I hit another limitation.</p><p>My vocabulary is narrower. I often use simpler words. Sometimes these words are technically correct, but semantically ambiguous.</p><p>For humans, this is manageable. For AI, it creates drift.</p><p>At the same time, this constraint forces something useful:</p><p>More precision. Less noise.</p><p>And with AI, precision matters more than expressiveness.</p><h3>Don’t start with code</h3><p>Most people approach AI like this:</p><p>&gt; “Here’s the idea — build it.”</p><p>This is the fastest way to get something wrong.</p><p>The approach that actually works is different:</p><ol><li>Define terminology</li><li>Lock naming conventions</li><li>Describe entities and relationships</li><li>Let AI ask questions</li><li>Refine understanding</li><li>Only then generate code</li></ol><p>This feels slower. It isn’t. AI can write code in seconds. Refactoring unclear code takes hours.</p><p>The key shift is simple:</p><p>Don’t ask AI to build. Ask it to interrogate you.</p><p>&gt; “Ask me everything you need to fully define this system.”</p><p>This forces you to externalise assumptions — things that are obvious to you but invisible to AI.</p><p>This is not new. This is what business analysts have always done. What changes now is the leverage. AI handles a large part of implementation. The bottleneck moves to understanding and definition.</p><p>Which means:</p><p>People who can clearly describe problems gain more influence than people who can just implement solutions. Not because they write code. Because they define it. AI doesn’t remove complexity. It shifts it.</p><p>From code → to thinking</p><p>From implementation → to communication</p><p>The developers who benefit the most are not the fastest coders. They are the ones who can clearly describe what needs to be built.</p><p>Naming things is still hard. Now it’s also critical.</p><p>Looking for your comments, no matter if you agree or don’t! <br>Let’s connect on <a href="https://www.linkedin.com/in/dmytro-sirant/"><strong>LinkedIn</strong></a><strong> !</strong></p><img src="https://medium.com/_/stat?event=post.clientViewed&referrerSource=full_rss&postId=595591b76375" width="1" height="1" alt="">]]></content:encoded>
        </item>
        <item>
            <title><![CDATA[How I Overlooked the Problem and Shot Myself in the Foot]]></title>
            <link>https://medium.com/@sirantd/how-i-overlooked-the-problem-and-shot-myself-in-the-foot-06841414e1de?source=rss-cd0af9ee77ad------2</link>
            <guid isPermaLink="false">https://medium.com/p/06841414e1de</guid>
            <category><![CDATA[aws]]></category>
            <category><![CDATA[cloud-computing]]></category>
            <category><![CDATA[security]]></category>
            <category><![CDATA[migration]]></category>
            <category><![CDATA[aws-kms]]></category>
            <dc:creator><![CDATA[Dmytro Sirant]]></dc:creator>
            <pubDate>Fri, 07 Nov 2025 02:43:17 GMT</pubDate>
            <atom:updated>2025-11-07T02:43:17.514Z</atom:updated>
            <content:encoded><![CDATA[<figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*JS0cdvNUNEAXKFar-j1P7A.png" /></figure><h3>Migration Setup</h3><p>As part of my work as an AWS consultant, I’ve been doing migrations from IAM Users to SSO (yes, I know what you think, but better late than never). There’s a checklist I follow during such migrations, and one of the last stages is to keep IAM Users disabled (keys and web access) for about a month, just to ensure that everything works fine without issues. This approach proved helpful in the case of EKS clusters before AWS introduced the ability to manage cluster access via API instead of the legacy aws-auth configmap.</p><h3>The Missed Detail</h3><p>But there was one issue I kept overlooking until it finally caught up with me. Long story short: the migration from IAM to SSO was completed, and after the planned cooldown, IAM users were deleted. Some time later, I decided to upgrade the IaC with a new version of the <a href="https://github.com/terraform-aws-modules/terraform-aws-eks">terraform-aws-eks</a> module. The terraform plan showed expected changes, but during terraform apply I got an error stating that my SSO account had no permission to update the KMS key alias (a minor change due to improved naming conventions).</p><p>A quick check showed that the KMS key created by the previous version of the module had a neat, least-privilege key policy — with `kms:PutPolicy` permission granted only to the IAM user I’d used to create the EKS cluster with KMS envelope encryption. Ironically, that IAM user had been disabled for a month and deleted only a week earlier.</p><h3>False Sense of Victory</h3><p>My first thought was that it wasn’t a big deal — I’d just remove the current KMS key object from the Terraform state, let it create a new key, and associate it with the cluster. Sounded good. Even better, terraform plan and terraform applycompleted without errors. Problem solved!</p><p>Or so I thought.</p><p>After a few small tweaks, I ran another change and noticed that Terraform tried to update the EKS cluster resource again. The only difference was the KMS envelope encryption key association. Multiple runs, same behaviour - Terraform applied the change successfully, yet it wasn’t reflected in the cluster settings (it still used the old KMS key I had no access to).</p><p>A quick check of the documentation confirmed that <a href="https://docs.aws.amazon.com/eks/latest/userguide/envelope-encryption.html">the envelope encryption key</a> can’t be changed after creation. Fair enough. But why didn’t Terraform respect that? I’m not sure yet whether the AWS API isn’t returning a proper response code, or the Terraform AWS provider doesn’t handle it correctly.</p><h3>Recovery Attempt</h3><p>So, I needed recovery access to the KMS key somehow. How do you get maximum access permissions to an AWS account? Correct — use the root login. Unfortunately, even with root permissions, I couldn’t update the KMS key policy. Great for security, but I still needed to regain access to the key. Time to check if anyone else had faced the same issue, and within a few minutes the answer was clear — contact AWS Support.</p><p>I opened a support case, explained the problem, and provided the ARN of the KMS key along with the policy I wanted to attach. What could be simpler? The support response surprised me. I had to follow a specific set of instructions:</p><blockquote>To recover your unmanageable keys, please follow these steps:</blockquote><blockquote>1. Create 1 IAM user for every affected key using the following naming convention: kms_key_recovery_&lt;KEY_ID&gt; For Example. If the Key ID is “17e51010-cc0f-2268-bccd-2699f10c133a” , then the corresponding recovery user would be “kms_key_recovery_17e51010-cc0f-2268-bccd-2699f10c133a”. Create and attach the following IAM policy to the newly created users: { “Version”: “2012–10–17”, “Statement”: [ { “Effect”: “Allow”, “Action”: [ “kms:ListAliases” ], “Resource”: “*” } ] }</blockquote><blockquote>2. Once the recovery users have been created, please respond to this case with a list of the ARNs of the keys for which you have made recovery users and the respective users ARNs.</blockquote><blockquote>3. We will then contact you via the phone number listed in your account Contact Information to verify the One-Time Password listed above. Once verified, we will engage our internal KMS team to initiate recovery and update your case when complete. Your key recovery users will have the ability to modify the key policies of their respective keys. If the process was abandoned, we will inform you which users in your account still have access.</blockquote><p>Because it’s not a synchronous operation and the KMS key policy can be updated at any time within a 12-hour window, I was worried the cluster might enter a degraded state before I could apply the proper policy. Luckily, that didn’t happen. The AWS team just granted the required permissions to the key as an additional rule without dropping the existing ones. The next day, I received an update confirming the recovery procedure was complete, and I was able to attach the correct policy to the key.</p><h3>Lessons Learnt</h3><p>Now there’s a new step in my IAM-to-SSO migration checklist:</p><p><strong>☑ update KMS key policies before deleting IAM users</strong></p><p>Looking for your comments, suggestions and improvements and let’s connect on <a href="https://www.linkedin.com/in/dmytro-sirant/"><strong>LinkedIn</strong></a><strong> !</strong></p><img src="https://medium.com/_/stat?event=post.clientViewed&referrerSource=full_rss&postId=06841414e1de" width="1" height="1" alt="">]]></content:encoded>
        </item>
        <item>
            <title><![CDATA[AWS and Docker Hub Limits: Smart Strategies for April 2025 Changes]]></title>
            <link>https://medium.com/@sirantd/aws-and-docker-hub-limits-smart-strategies-for-april-2025-changes-42bd9295cad6?source=rss-cd0af9ee77ad------2</link>
            <guid isPermaLink="false">https://medium.com/p/42bd9295cad6</guid>
            <category><![CDATA[docker]]></category>
            <category><![CDATA[terraform]]></category>
            <category><![CDATA[kubernetes]]></category>
            <category><![CDATA[aws]]></category>
            <category><![CDATA[infrastructure]]></category>
            <dc:creator><![CDATA[Dmytro Sirant]]></dc:creator>
            <pubDate>Wed, 26 Feb 2025 09:06:17 GMT</pubDate>
            <atom:updated>2025-02-27T05:38:36.910Z</atom:updated>
            <content:encoded><![CDATA[<figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*_1owtTQhXalqmCN6aO5aZA.png" /></figure><h3>Problem</h3><p>Docker Hub has implemented rate limits on image pulls back on November 2, 2020, which made a lot of headaches for those who are using it in their environment. The following limits were introduced:</p><ul><li><strong>Anonymous Users</strong>: Limited to 100 pulls per 6 hours</li><li><strong>Authenticated Users</strong>: Limited to 200 pulls per 6 hours</li></ul><p>Many of us migrated to private repositories, quay.io, ghrc.io, registry.gitlab.com, and others. For some, it was enough to use an authentication token to pull from Docker Hub.</p><p><strong>Starting from April 1, 2025 there is a new limits:</strong></p><ul><li><strong>Personal (authenticated)</strong>: Limited to 100 pulls per hour</li><li><strong>Unauthenticated Users</strong>: Limited to 10 pulls per IPv4 address or IPv6 /64 subnet per hour</li></ul><h3>Solution</h3><p>Let’s see what we have at AWS to solve the problem with the rate limit and even more perks.</p><p>In November 2021, <a href="https://aws.amazon.com/about-aws/whats-new/2021/11/amazon-ecr-cache-repositories/">AWS introduced the pull-through cache feature for Amazon Elastic Container Registry (Amazon ECR)</a>, which can help us with the following problems:</p><ul><li>caching public and private images using an authentication token in your private ECR registry</li><li>speedup pulling from private ECR to your local services (ECS, EKS, Lambdas, etc.)</li><li>lifecycle policy to keep only the required number of the latest tags</li><li>security scanning of images during pull</li><li>a single place to update your token in case of rotation or expiration (e.g. Gitlab do not allow you to create tokens with an expiration date longer than one year). Just imagine you need to go through all your credentials in all K8s clusters one per year to update tokens.</li></ul><p>So, let’s dive into the implementation of this feature. To make it work, you need to create:</p><ol><li>a <strong>pull-through cache rule</strong>, where you define the prefix you want to use in your registry and the destination registry where you forward all requests</li><li><strong>secrets manager credentials</strong> you want to use to authenticate to the destination registry</li><li>a <strong>pull-through cache repository template</strong> if you want to have a lifecycle policy, security scans</li></ol><p>and some other features, which are optional at this stage.</p><p>As you can see, you are not creating the repository itself. It will be created automatically during the first pull-through cache request using default settings or a template if you have assigned one.</p><p>If I created a rule with the prefix `<strong>dockerhub</strong>` in the `<strong>us-east-1</strong>` region and say to forward my pull requests to the `<strong>registry-1.docker.io</strong>`, I can use the following command:</p><pre># direct pull from Docker Hub<br>docker pull timberio/vector:0.45.0-alpine<br># pull through ECR<br>docker pull 123456789012.dkr.ecr.us-east-1.amazonaws.com/dockerhub/timberio/vector:0.45.0-alpine</pre><p>As you can see, I only added my repository and `<strong>dockerhub</strong>` as prefix, and then the original image to pull. It’s what you need to update in manifests later to pull images through ECR, which is pretty straightforward.</p><h3>Automation</h3><p>Sounds easy, but why not make it using Terraform (with a <a href="http://terragrunt.gruntwork.io/">Terragrunt</a> wrapper)? I created a simple Terraform module which helps you create multiple pull-through rules from the basic YAML file with the following structure:</p><pre>dockerhub:<br>  registry: registry-1.docker.io<br>  username: user<br>  accessToken: token<br>gitlab:<br>  registry: registry.gitlab.com<br>  username: user<br>  accessToken: token</pre><p>and module will create two rules with prefixes: `<strong>dockerhub</strong>` and `<strong>gitlab</strong>` , secrets manager credentials with `<strong>username</strong>` and `<strong>accessToken</strong>`, and basic lifecycle policy to keep only 3 latest images in cache (makes sense to extend the YAML with per-registry settings but I have no time for it now).</p><p>Because I’m using Terragrunt I can encrypt this file with <a href="https://github.com/getsops/sops">Sops</a> and commit to the repository safely, so my `<strong>terragrunt.hcl</strong>` is looking like that:</p><pre>terraform {<br>  source = &quot;${get_parent_terragrunt_dir()}modules/ecr-pullthrough&quot;<br>}<br><br>include &quot;root&quot; {<br>  path = find_in_parent_folders(&quot;root.hcl&quot;)<br>  expose = true<br>}<br><br>locals {<br>  registries = yamldecode(sops_decrypt_file(&quot;${get_terragrunt_dir()}/secrets.enc.yaml&quot;))<br>}<br><br>inputs = {<br>  registries = local.registries<br>  tags = merge(include.root.locals.default_tags)<br>}</pre><h4><strong>You can find my module on Github </strong><a href="https://github.com/opsworks-co/ecr-pull-through"><strong>https://github.com/opsworks-co/ecr-pull-through</strong></a></h4><blockquote>If you are using <a href="https://github.com/terraform-aws-modules/terraform-aws-eks">https://github.com/terraform-aws-modules/terraform-aws-eks</a> module to create your cluster and using in-built role it is creating for the worker nodes, don’t forget to extend it with a new policy to allow `<strong>ecr:BatchImportUpstreamImage</strong>` because by default it attaches managed policy `<strong>arn:aws:iam::aws:policy/AmazonEC2ContainerRegistryPullOnly</strong>` which doesn’t allow pull-through cache.</blockquote><p>Looking for your comments, suggestions and improvements and let’s connect on <a href="https://www.linkedin.com/in/dmytro-sirant/"><strong>LinkedIn</strong></a><strong> !</strong></p><img src="https://medium.com/_/stat?event=post.clientViewed&referrerSource=full_rss&postId=42bd9295cad6" width="1" height="1" alt="">]]></content:encoded>
        </item>
        <item>
            <title><![CDATA[Choosing the Best Terraform Approach for IAM Policies: Heredoc vs jsonencode() vs…]]></title>
            <link>https://medium.com/@sirantd/choosing-the-best-terraform-approach-for-iam-policies-heredoc-vs-jsonencode-vs-8e4948f8e756?source=rss-cd0af9ee77ad------2</link>
            <guid isPermaLink="false">https://medium.com/p/8e4948f8e756</guid>
            <category><![CDATA[heredoc]]></category>
            <category><![CDATA[aws]]></category>
            <category><![CDATA[devops]]></category>
            <category><![CDATA[terraform]]></category>
            <category><![CDATA[iac]]></category>
            <dc:creator><![CDATA[Dmytro Sirant]]></dc:creator>
            <pubDate>Mon, 06 Jan 2025 05:41:05 GMT</pubDate>
            <atom:updated>2025-01-06T05:41:05.098Z</atom:updated>
            <content:encoded><![CDATA[<h3>Choosing the Best Terraform Approach for IAM Policies: Heredoc vs jsonencode() vs aws_iam_policy_document</h3><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*kOI-7IOETV2DIFtAPlWkAQ.png" /><figcaption>Heredoc vs jsonencode() vs aws_iam_policy_document</figcaption></figure><p>As with many things in IT, there are often multiple ways to achieve the same goal. Sometimes the differences are subtle and come down to personal habits, but most of the time, each option has its own advantages and trade-offs. While working on multiple Terraform projects, I’ve noticed that engineers are often unaware of these nuances, leading to inconsistent approaches being copy-pasted across different parts of a project. Let’s fix that and dive deeper into the options available and compare they advantages and trade-offs.</p><h3>1. Resource aws_iam_policy and EOF</h3><pre>resource &quot;aws_iam_policy&quot; &quot;policy&quot; {<br>  name        = &quot;test_policy&quot;<br>  path        = &quot;/&quot;<br>  description = &quot;My test policy&quot;<br>  policy = &lt;&lt;EOF<br>{<br>  &quot;Version&quot; = &quot;2012-10-17&quot;<br>  &quot;Statement&quot; = [<br>    {<br>      &quot;Action&quot; = [<br>        &quot;ec2:Describe*&quot;,<br>      ]<br>      &quot;Effect&quot;   = &quot;Allow&quot;<br>      &quot;Resource&quot; = &quot;*&quot;<br>    },<br>  ]<br>}<br>EOF</pre><p>You might seen this way of declaring policy in your code often. It uses the well known approach (I didn’t know that it even has name — heredoc) to pass data to variable which ends with user-defined delimiter (in my example it&#39;s EOF but could be anything you like)</p><p>In my opinion, this is the worst possible way to create a policy due to several reasons:</p><ul><li>❌ difficult to format JSON code</li><li>❌ hard to read when mixing Terraform and JSON code</li><li>❌ no syntax highlighting for JSON code</li><li>✅ Terraform validates if correct JSON passed</li><li>✅ you can use variables as ${var.some_variable} in your policy</li><li>❌ no policy checks before applying</li></ul><h3>2. Resource aws_iam_policy and jsonencode()</h3><pre>resource &quot;aws_iam_policy&quot; &quot;policy&quot; {<br>  name        = &quot;test_policy&quot;<br>  path        = &quot;/&quot;<br>  description = &quot;My test policy&quot;<br>  policy = jsonencode({<br>      Version = &quot;2012-10-17&quot;<br>      Statement = [<br>        {<br>          Action = [<br>            &quot;ec2:Describe*&quot;,<br>          ]<br>          Effect   = &quot;Allow&quot;<br>          Resource = &quot;*&quot;<br>        },<br>      ]<br>    })<br>}</pre><p>This example from the Terraform documentation is an improvement because it resolves the issues present in the first approach. It uses the built-in jsonencode() function to convert Terraform code into JSON, ensuring a consistent coding style. It also helps avoid common problems like formatting errors, whitespace issues, and other JSON-specific challenges.</p><p>While this approach is much better, it’s still not perfect:</p><ul><li>✅ the JSON block is automatically generated for you</li><li>✅ the JSON block is easy to work with</li><li>✅ Terraform code highlighting is maintained, as there’s no direct JSON code</li><li>✅ jsonencode() validates your Terraform syntax and generates correct JSON</li><li>✅ you can use variables as ${var.some_variable} in your policy</li><li>❌ no policy check before applying</li></ul><h3>3. Data Source aws_iam_policy_document</h3><p>In the two previous examples I highlighted the problem that policy will not be checked before apply. Creating a valid policy requires two components:</p><ul><li>having valid JSON syntax</li><li>having valid policy syntax</li></ul><p>It could be situation, when you trying to add some block which is proper from the JSON syntax, but not expected in the policy you are working on, e.g. AWSPrincipal:</p><pre>resource &quot;aws_iam_policy&quot; &quot;policy&quot; {<br>    name = &quot;test_policy&quot;<br>    path = &quot;/&quot;<br>    description = &quot;My test policy&quot;<br>    policy = jsonencode({<br>        Version = &quot;2012-10-17&quot;<br>        Statement = [<br>        {<br>            Action = [<br>                &quot;ec2:Describe*&quot;,<br>            ]<br>            AWSPrincipal = {<br>                AWS = [<br>                    &quot;arn:aws:iam::AWS-account-ID:user/user-name-1&quot;,<br>                    &quot;arn:aws:iam::AWS-account-ID:user/user-name-2&quot;<br>                ]<br>            }<br>            Effect = &quot;Allow&quot;<br>            Resource = &quot;*&quot;<br>        },<br>        ]<br>    })<br>}<br><br>resource &quot;aws_iam_policy&quot; &quot;example&quot; {<br>    name = &quot;example_policy&quot;<br>    path = &quot;/&quot;<br>    policy = aws_iam_policy.policy.arn<br>}</pre><p>Both heredoc and jsonencode() will accept this block because it&#39;s valid JSON, you will not have errors with terraform validate and terraform plan will show you changes in your policy document you want to apply but then policy update failed with error:</p><pre>aws_iam_policy.policy: Creating...<br>╷<br>│ Error: creating IAM Policy (test_policy): operation error IAM: CreatePolicy,<br>│ https response error StatusCode: 400, RequestID: aaaa-bbbb-cccc-8782-b2dda037f135, <br>│ MalformedPolicyDocument: Syntax errors in policy.<br>│ <br>│   with aws_iam_policy.policy,<br>│   on terraform.tf line 24, in resource &quot;aws_iam_policy&quot; &quot;policy&quot;:<br>│   24: resource &quot;aws_iam_policy&quot; &quot;policy&quot; {<br>│ <br>╵</pre><p>Using aws_iam_policy_document will help you avoid such problems because of error will be rise with terraform validate and terraform plan. So, if I run the following code:</p><pre>data &quot;aws_iam_policy_document&quot; &quot;example&quot; {<br>    statement {<br>        sid = &quot;1&quot;<br>        AWSPrincipal {<br>            type = &quot;AWS&quot;<br>            identifiers = [<br>                &quot;arn:aws:iam::AWS-account-ID:user/user-name-1&quot;,<br>                &quot;arn:aws:iam::AWS-account-ID:user/user-name-2&quot;<br>            ]<br>        }<br>        actions = [<br>            &quot;iam:ListVirtualMFADevices&quot;,<br>            &quot;iam:GetAccountPasswordPolicy&quot;<br>        ]<br>        resources = [<br>            &quot;*&quot;,<br>        ]<br>    }<br>}<br><br>resource &quot;aws_iam_policy&quot; &quot;example&quot; {<br>    name = &quot;example_policy&quot;<br>    path = &quot;/&quot;<br>    policy = data.aws_iam_policy_document.example.json<br>}</pre><p>I’ll have the error with terraform validate:</p><pre>❯ terraform validate<br>╷<br>│ Error: Unsupported block type<br>│ <br>│   on terraform.tf line 5, in data &quot;aws_iam_policy_document&quot; &quot;example&quot;:<br>│    5:     AWSPrincipal {<br>│ <br>│ Blocks of type &quot;AWSPrincipal&quot; are not expected here.</pre><p>If we fix error and have proper syntax for aws_iam_policy_document it will pass terraform validate but fails later during terraform apply:</p><pre>aws_iam_policy.example: Creating...<br>╷<br>│ Error: creating IAM Policy (example_policy): operation error IAM: CreatePolicy, <br>│ https response error StatusCode: 400, RequestID: aaaa-bbbb-cccc-b32b-2a316c5ff8e6, <br>│ MalformedPolicyDocument: Policy document should not specify a principal.<br>│ <br>│   with aws_iam_policy.example,<br>│   on terraform.tf line 26, in resource &quot;aws_iam_policy&quot; &quot;example&quot;:<br>│   26: resource &quot;aws_iam_policy&quot; &quot;example&quot; {<br>│</pre><p>This approach adds some more checks:</p><ul><li>✅ the JSON block is automatically generated for you</li><li>✅ the JSON block is easy to work with</li><li>✅ Terraform code highlighting is maintained, as there’s no direct JSON code</li><li>✅ jsonencode() validates your Terraform syntax and generates correct JSON</li><li>✅ you can use variables as ${var.some_variable} in your policy</li><li>✅/❌ additional attributes validation, but still no policy check before applying</li></ul><h3>Summary</h3><p>I see no reason to use the heredoc style to define your policy. It offers no advantages over the other two available options and has multiple trade-offs.</p><p>While jsonencode() and the Data Source aws_iam_policy_document provide additional checks for JSON syntax and policy syntax, they will not protect you from errors in your policy during the apply stage. Be cautious when working with policies, and refer to the documentation to ensure you are generating a proper policy. Nevertheless, your code will be easier to read and more structured if you follow either of these two approaches instead of using heredoc.</p><p><em>Looking forward to your feedback in the comments!</em></p><p>Let’s connect on <a href="https://www.linkedin.com/in/dmytro-sirant/"><strong>LinkedIn</strong></a><strong> and </strong><a href="https://x.com/sirantd"><strong>X (Twitter)</strong></a><strong>!</strong></p><img src="https://medium.com/_/stat?event=post.clientViewed&referrerSource=full_rss&postId=8e4948f8e756" width="1" height="1" alt="">]]></content:encoded>
        </item>
        <item>
            <title><![CDATA[Looking back, year 2024]]></title>
            <link>https://medium.com/@sirantd/looking-back-year-2024-007420775f8e?source=rss-cd0af9ee77ad------2</link>
            <guid isPermaLink="false">https://medium.com/p/007420775f8e</guid>
            <category><![CDATA[reflections]]></category>
            <category><![CDATA[2024]]></category>
            <category><![CDATA[networking]]></category>
            <category><![CDATA[self]]></category>
            <dc:creator><![CDATA[Dmytro Sirant]]></dc:creator>
            <pubDate>Tue, 24 Dec 2024 03:20:51 GMT</pubDate>
            <atom:updated>2024-12-24T03:20:51.361Z</atom:updated>
            <content:encoded><![CDATA[<figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*szGyjpvATGbZUfwVuJ5C2A.png" /></figure><p>I’m one of those people who always think that everyone else around them is more productive, more successful, more focused. As a result, I suffer from imposter syndrome and can easily fall into procrastination. There’s nothing wrong with having bright people around you who inspire you to improve, but it’s hard to compete with all of them all the time. This year, I want to step back and reflect on 2024 — perhaps it wasn’t such a bad year for my professional and personal development. There’s no particular order in how I’ve listed these achievements, just an attempt to recall them month by month.</p><h3><strong>Public Speaking</strong></h3><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*QnWRSl0x621ZVlGlkcY-IQ.jpeg" /><figcaption>AWS User Group Brisbane Meetup, October 2024</figcaption></figure><p>I still can’t believe it happened, but this year I started with a presentation at the AWS User Group Brisbane in February. It wasn’t an easy step for me, but it opened a new chapter in my life. There were five public speaking activities this year:</p><ul><li><a href="https://www.linkedin.com/posts/dmytro-sirant_aws-awscommunity-usergroup-activity-7166270852232044544-ye2o?utm_source=share&amp;utm_medium=member_desktop"><strong>“Navigating AWS: A Guide to Seeking Support”</strong></a><strong> </strong>at<strong> </strong>AWS User Group Brisbane in February.</li><li><a href="https://www.linkedin.com/posts/dmytro-sirant_aws-awscommunity-awsambassador-activity-7186981143273959424-0vA3?utm_source=share&amp;utm_medium=member_desktop"><strong>“AWS Summit Sydney re:Cap Panel Discussion”</strong></a><strong> </strong>at AWS User Group Brisbane in April.</li><li><a href="https://www.linkedin.com/posts/dmytro-sirant_brisbane-serverless-meetup-in-person-and-activity-7236255170764922880-2a_A?utm_source=share&amp;utm_medium=member_desktop"><strong>“AWS Serverless — Not Lambda Alone”</strong></a><strong> </strong>at Brisbane Serverless Meetup in September.</li><li><a href="https://www.linkedin.com/posts/dmytro-sirant_aws-awsusergroupbrisbane-meetup-activity-7252846713906868224-2ZyR?utm_source=share&amp;utm_medium=member_desktop"><strong>“So You’ve Inherited a Cloud Environment, Episode 2”</strong></a><strong> </strong>at AWS User Group Brisbane in September.</li><li><a href="https://www.linkedin.com/posts/dmytro-sirant_aws-usergroup-awsusergroupgoldcoast-activity-7267892852708261889-P2GC?utm_source=share&amp;utm_medium=member_desktop"><strong>“Value of Certifications in IT Panel Discussion”</strong></a><strong> </strong>at AWS User Group Gold Coast in November.</li></ul><h3><strong>Education</strong></h3><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*dgOXdcjYYXS5TCgEs1COOQ.jpeg" /><figcaption>Graduation ceremony, TAFE 2024</figcaption></figure><p>This year, I graduated from TAFE with two diplomas: <a href="https://www.linkedin.com/posts/dmytro-sirant_ive-been-inactive-for-the-last-few-weeks-activity-7208377091237560320-GtEH?utm_source=share&amp;utm_medium=member_desktop"><strong>Diploma of Business</strong> and <strong>Diploma of Leadership and Management</strong></a>. When I enrolled in the course, I didn’t expect it to be so impactful. For some people, going back to school at 44 is challenging, but I found studying this time far more interesting compared to when I was 18. My teacher and fellow students made this year unforgettable and planted seeds of change in my mindset.</p><h3><strong>Content Creation</strong></h3><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*HM5c9ASvcfORZkmMKewimg.png" /><figcaption>Screenshot from the YouTube channel</figcaption></figure><p>Throughout the year, I continued working on content for <a href="https://www.youtube.com/channel/UCF36e8jNc6e3ge-y4hxx4_g"><strong>YouTube channel</strong></a>:</p><ul><li><a href="https://youtu.be/2CcJ2gFr1Bc"><strong>Collecting Logs and Metrics with Vector (Terraform + EKS + S3 + OpenSearch)</strong></a></li><li><a href="https://youtu.be/w5-kluapJTM"><strong>Типові помилки в архітектурі даних та як їх уникнути</strong></a> (in Ukrainian for the AWS User Group Kyiv conference)</li><li><a href="https://youtu.be/xCV_GvKqzP4"><strong>Value of IT Certification in 2025 for Career Development: Panel Discussion</strong></a></li></ul><p>Along with other videos made by my colleagues, it reached 403 subscribers organically and continue to grow.</p><h3><strong>Writing</strong></h3><p>I still feel uncomfortable in front of the camera, so I continue writing articles in both Ukrainian and English. In total, I’ve written more than 10 articles this year across several platforms:</p><ul><li><a href="https://dou.ua/users/dmytro-sirant/articles/">DOU Articles</a></li><li><a href="https://sirantd.com/">Medium Blog</a></li><li><a href="https://www.opsworks.co/blog">Company Blog</a></li></ul><h3><strong>Certifications</strong></h3><figure><img alt="" src="https://cdn-images-1.medium.com/max/777/1*PbIyz3sM9oHR3t82Fc8yOQ.png" /><figcaption>15x AWS certificates I got</figcaption></figure><p>This year, I completed my certification journey and achieved <strong>15 AWS active certifications</strong>. It was challenging but fascinating to learn, and I’m not planning to stop. I’m looking for new challenges — if you have any suggestions, please drop me a comment!</p><h3><strong>Personal Development</strong></h3><p>As a perfectionist by nature, I’m not satisfied with my English. Two months ago, I started seeing a speech therapist to work on my accent. It’s harder than I expected, but I’m committed to improving!</p><p>There are also other activities worth mentioning, such as attending conferences, contributing to open-source projects, and helping clients but to be honest, I didn’t achieve everything I planned this year, but that’s a story for another time. Stay tuned!</p><p>Let’s connect on <a href="https://www.linkedin.com/in/dmytro-sirant/"><strong>LinkedIn</strong></a><strong> and </strong><a href="https://x.com/sirantd"><strong>X (Twitter)</strong></a><strong>!</strong></p><img src="https://medium.com/_/stat?event=post.clientViewed&referrerSource=full_rss&postId=007420775f8e" width="1" height="1" alt="">]]></content:encoded>
        </item>
        <item>
            <title><![CDATA[From Ukraine to Australia: My Path to Building the AWS User Group Gold Coast]]></title>
            <link>https://medium.com/@sirantd/from-ukraine-to-australia-my-path-to-building-the-aws-user-group-gold-coast-5204c5eb2012?source=rss-cd0af9ee77ad------2</link>
            <guid isPermaLink="false">https://medium.com/p/5204c5eb2012</guid>
            <category><![CDATA[aws]]></category>
            <category><![CDATA[awscommunitybuilder]]></category>
            <category><![CDATA[gold-coast]]></category>
            <category><![CDATA[networking]]></category>
            <category><![CDATA[community]]></category>
            <dc:creator><![CDATA[Dmytro Sirant]]></dc:creator>
            <pubDate>Mon, 23 Dec 2024 02:05:39 GMT</pubDate>
            <atom:updated>2024-12-23T02:05:39.447Z</atom:updated>
            <content:encoded><![CDATA[<figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*l-LhWePGPAdgH55TnSsZqg.png" /><figcaption>AWS User Group Gold Coast, Australia</figcaption></figure><p>It’s the first year that I’m running the AWS User Group Gold Coast, but it took me longer to prepare emotionally and finally realize that I was ready. Let me start a few years earlier to make the story complete.</p><p>I arrived on the Gold Coast, Australia, in March 2022 to find a safe haven for my family because of the full-scale war russia started in Ukraine. COVID restrictions had just begun to ease, and I was invited as an AWS partner representative to the AWS Summit 2022 After Party in Brisbane. I met interesting people there and discovered AWS User Group events in Brisbane and on the Gold Coast. Because of COVID, many events were conducted online, which is not the best experience you can have. No surprise that the User Group Gold Coast wasn’t active. On the flip side, the AWS User Group Brisbane recovered well, and I attended the events they held every one or two months. Brisbane is about 70 km from where I live, so it wasn’t a long trip to have a chat with people who share your interests. For a person who is kind of introverted and not a native English speaker, it was really hard mentally to become part of the community.</p><p>It took about a year and a half and enormous support from the community to finally come to the stage at one of the events in early 2024. I was terrified, couldn’t sleep for a few days before the presentation, but I had a great feeling afterward. In June 2024, I finished my studies and decided I was ready for new goals.</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*BziHNVX-7Yp9AEW9bp9xmQ.jpeg" /><figcaption>First time presenting on a stage, 22 February 2024</figcaption></figure><p>Long story short, I got support from <a href="https://www.linkedin.com/in/alan-blockley/">Alan Blockley</a> in organising the User Group on the Gold Coast and approached <a href="https://www.linkedin.com/in/dalton/">Steve Dalton</a> with the idea to revive it. It took some time to find the venue, organize speakers, and create visuals, but on the 26th of September 2024, the group held its first event. More than 30 people attended, we had great presentations, and the event lasted four hours instead of the planned two, simply because everyone was so happy to chat.</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*jZZgyGoIyy0XW6u_ttw9tw.jpeg" /><figcaption>26th of September 2024, first AWS User Group meetup after restart</figcaption></figure><p>Seeing the demand for such events, both for seasoned professionals and students just starting their journey in the IT world, it was clear that this group was needed for the Gold Coast community. I’m still looking for the best format by talking to people, collecting feedback, and learning from other communities. As a result, the idea for the second event, hosted on the 28th of November 2024, emerged from a LinkedIn conversation about the value of IT certifications. This led to a great panel discussion. Another 25 people attended, and the recordings were <a href="https://www.youtube.com/playlist?list=PLiyuz9BeSyCCIrpJ5A5_N4VRd5s9quQbT">published on YouTube</a> for those unable to join in person.</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*JwZ1hsXj-eFQO1yBgp21iA.jpeg" /></figure><p>December is a busy month because of AWS re:Invent, finishing tasks, reflecting on the year, and making plans for the next. We are currently preparing for our next meeting on the 16th of January — it’s going to be a great re:Cap of re:Invent 2024 with many interesting people coming to the Gold Coast. <a href="https://www.meetup.com/awsgoldcoast/events/305111724/?utm_medium=referral&amp;utm_campaign=share-btn_savedevents_share_modal&amp;utm_source=link">Don’t miss it!</a></p><figure><img alt="" src="https://cdn-images-1.medium.com/max/960/1*EsMYORm8VxT_hTDAv8kIvQ.png" /><figcaption>16th of January 2025 AWS re:Invent re:Cap at Gold Coast</figcaption></figure><p>If you have a story to share with the community — don’t be afraid! Get in touch with me, and I’ll do my best to organise it and provide help if you need it!</p><p>Follow me on Medium, and let’s connect on <a href="https://www.linkedin.com/in/dmytro-sirant/"><strong>LinkedIn</strong></a><strong> and </strong><a href="https://x.com/sirantd"><strong>X (Twitter)</strong></a><strong>!</strong></p><img src="https://medium.com/_/stat?event=post.clientViewed&referrerSource=full_rss&postId=5204c5eb2012" width="1" height="1" alt="">]]></content:encoded>
        </item>
        <item>
            <title><![CDATA[LinkedIn Networking: Quantity vs Quality]]></title>
            <link>https://medium.com/@sirantd/linkedin-networking-quantity-vs-quality-7e7a6029a204?source=rss-cd0af9ee77ad------2</link>
            <guid isPermaLink="false">https://medium.com/p/7e7a6029a204</guid>
            <category><![CDATA[sales]]></category>
            <category><![CDATA[networking]]></category>
            <category><![CDATA[social-media-marketing]]></category>
            <category><![CDATA[linkedin]]></category>
            <category><![CDATA[social-media]]></category>
            <dc:creator><![CDATA[Dmytro Sirant]]></dc:creator>
            <pubDate>Mon, 09 Dec 2024 00:10:17 GMT</pubDate>
            <atom:updated>2024-12-09T00:10:17.092Z</atom:updated>
            <content:encoded><![CDATA[<figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*1AWRVNQqUnUxFVaS_iNhIQ.png" /></figure><p>Three years ago, I had around 6,000 connections on LinkedIn. Were they meaningful? Not really. Did it please my ego? Absolutely.</p><p>Let me explain how I got to this point. Back in 2019, having the magic word <em>“DevOps”</em> in your title made you a sought-after prize for recruiters. They sent connection requests, and I accepted them unconditionally. I wasn’t actively looking for opportunities or had an “Open to work” status, but I figured it wouldn’t hurt to be connected — just in case.</p><p>One day, I realized my newsfeed was irrelevant. The people I was connected with didn’t align with my interests or, worse, were just bots. That was a turning point. I decided to clean up my network.</p><p>It wasn’t easy to remove 3,500 connections. LinkedIn doesn’t provide automation tools for this — or at least I couldn’t find them. However, it became my go-to task for procrastination. I created filters, reviewed search results, revoked connections, repeated the process, hit LinkedIn’s rate limits, got flagged for using automation tools (which I wasn’t), and had to convince LinkedIn it was just me manually clicking through the interface. After a while, I ended up with a much smaller, cleaner network — 2,500 connections with people whose content I genuinely cared about.</p><p>Since then, several things have changed. First, I started reviewing connection requests more carefully, ignoring those that weren’t relevant to me. Occasionally, I send connection requests myself to people I follow, meet in person, or collaborate with on projects.</p><p>Second, my role changed to Chief Technology Officer, putting me on the radar of salespeople, business development managers, and others who believe all they need to succeed is an automated LinkedIn sequence targeting CxOs, business owners, and decision-makers. Frankly, I find these approaches more annoying than recruiters (sorry to my sales friends who take a more thoughtful approach). The tactics some salespeople use to grab attention — like using a CEO’s profile in automated sequences, pretending to be interested in my posts, or making uninformed assumptions about my company’s needs — can be exhausting.</p><p>When I accept a connection, their behavior generally falls into three categories:</p><p>1. <strong>Immediate Sales Pitch:</strong> They start pitching right away. Ignoring them doesn’t work; follow-ups come like clockwork the next day or after another fixed period.</p><p>2. <strong>Polite Intro, Then Sales Pitch:</strong> They try to build rapport with general phrases or feigned interest in me, but after two or three messages, they revert to default sales behavior.</p><p>3. <strong>Passive Content Sharing (The Best Approach):</strong> They connect and focus on sharing content about their product or service. Since LinkedIn highlights content from my network, I may engage with posts that resonate, which could eventually lead to business opportunities.</p><p>For those in the first group, I revoke the connection without interaction. For the second group, I sometimes give feedback about my expectations. If their approach doesn’t improve, I remove them too.</p><p>As for myself, I follow the same principles — I never send sales pitches to my network, avoid automated sequences, and always respond personally to messages. I value my time and prefer genuine interactions over bots.</p><p><strong>I enjoy growing my network, but I’m not willing to compromise on its quality. If this resonates with you, let’s connect!</strong></p><p>Follow me on Medium, and let’s connect on <a href="https://www.linkedin.com/in/dmytro-sirant/"><strong>LinkedIn</strong></a><strong> and </strong><a href="https://x.com/sirantd"><strong>X (Twitter)</strong></a><strong>!</strong></p><img src="https://medium.com/_/stat?event=post.clientViewed&referrerSource=full_rss&postId=7e7a6029a204" width="1" height="1" alt="">]]></content:encoded>
        </item>
        <item>
            <title><![CDATA[A 30-Year Perspective: The New Realities of Studying IT Today]]></title>
            <link>https://medium.com/@sirantd/a-30-year-perspective-the-new-realities-of-studying-it-today-2e2bdc06eb02?source=rss-cd0af9ee77ad------2</link>
            <guid isPermaLink="false">https://medium.com/p/2e2bdc06eb02</guid>
            <category><![CDATA[information-technology]]></category>
            <category><![CDATA[learning]]></category>
            <category><![CDATA[students]]></category>
            <category><![CDATA[reflections]]></category>
            <category><![CDATA[knowledge]]></category>
            <dc:creator><![CDATA[Dmytro Sirant]]></dc:creator>
            <pubDate>Thu, 28 Nov 2024 00:21:47 GMT</pubDate>
            <atom:updated>2024-11-28T00:21:47.528Z</atom:updated>
            <content:encoded><![CDATA[<figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*K2vNIZ5wmBlPIPvxmmc_4g.png" /></figure><p>I find it challenging to explain complex concepts to someone without any foundational knowledge in the field. Analogies can sometimes help, offering a way to present these concepts in a more relatable manner using vocabulary that resonates with the listener. This approach works well when speaking to non-technical businesspeople and trying to help them understand an idea. However, explaining something to a future IT professional requires a different level of detail and understanding.</p><p>As someone without a formal IT education, I often find it difficult to explain how I learned something — what steps and in what order someone else would need to take to achieve the same results. I’ve acquired my knowledge throughout my life, but I can’t distill it into concise learning materials that would help someone become a professional in a year or two. There’s nothing in my experience that could be described as “condensed wisdom” because my knowledge consists of many interconnected pieces. A good example is when I show the steps I performed during troubleshooting, and my colleague wonders: “How did you come straight to this hypothesis and skip all the other steps?” This is hard to teach without being misleading because, next time, I might not skip those steps. I’ve always envied those with a degree in Computer Science, who have a structured learning program. Occasionally, I feel gaps in my knowledge and contemplate enrolling in a CS course to finally address this lingering sense of incompleteness.</p><p>I got my first computer from my parents when I was in Grade 8, back in 1993. It was an “Electronica MC 1502”, equipped with an extra memory cartridge and a cartridge for BASIC programming. I also had external 5.25” floppy disk drives and a few disks with DOS. With no games to play, I spent my time experimenting with commands and writing my own programs in BASIC. Back then, living in an industrial (non-capital) city in Ukraine, I had no Internet, no local IT community, and almost no books or magazines on the subject.</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*12QmLDOu652Gmf0mYfyC8g.jpeg" /></figure><p>When you lack access to structured learning, you’re forced to build your own model of a system based on its reactions. I later learned that this is called the “black box” approach. Scientifically, studying a “black box” involves understanding a system or process without direct access to its internal workings, relying instead on observation, experimentation, and inference. I made countless mistakes, but over time, I managed to establish a foundation upon which I could build future layers of knowledge. As I acquired new insights, I often revisited and adjusted the basic system I had constructed in my mind.</p><p>I’ve tried to find a good analogy for the process of learning. One that resonates with me is the jigsaw puzzle. Most would agree that completing a puzzle becomes progressively easier as more pieces are connected. With more blocks in place, it’s easier to identify and connect additional pieces based on shared colors or shapes. In this analogy, formal education provides those foundational corner and edge pieces, making it easier to build out the entire picture later.</p><p>However, while this analogy works well for understanding basic concepts as a 2D structure, real-life learning involves not only interconnections in 2D but also many levels of abstraction, resembling a 3D structure. Let me explain using an example: on a 2D layer, we might have hardware, networking, and operating systems. These domains are interconnected but can still be learned as separate subjects. Each subsequent layer is usually built on top of multiple underlying layers. Building on this, higher abstraction layers include virtualization, IP protocols, kernels, and filesystems. Above those, we have containers, routing protocols, encryption, server applications, and so on. In my mind, it resembles a pyramid built of jigsaw puzzle pieces.</p><p>When I attended driving school 30 years ago, we were taught how to fix an engine on the side of the road if it broke down. Car owners were expected to know their vehicles inside out and often spent much of their time in the garage making repairs. I recall the box of spare parts and tools we always carried on family trips. Today, drivers are only expected to understand basic operational principles: turn indicators, windscreen wipers, and headlamps (though, regrettably, turn indicators often seem underutilized!). This shift is due to several factors: improved quality, automation, simplification, and reliance on specialized professionals for maintenance.</p><p>In IT, the pace of development is even faster. Many abstractions have matured, providing excellent isolation from lower levels. I’ve observed many people starting their IT careers at the top abstraction layers and considering themselves professionals because they’ve mastered a specific technology. For instance, I’ve encountered cloud engineers with no understanding of networking, Linux, or Windows basics. Some believe that being a DevOps engineer is solely about CI/CD pipelines and doesn’t require knowledge of programming languages or scripting in Bash or PowerShell. While my journey has been different — starting from the bottom layers and working upward — I still believe that those beginning with higher abstractions should make an effort to understand the underlying layers thoroughly and even explore adjacent areas.</p><p>I worry about a world where someone with just a six-month course uses generative AI to develop software for self-driving cars or surgical robots. These technologies demand a depth of understanding that cannot be replaced by surface-level expertise.</p><blockquote>The thoughts above are my opinion, and I’d appreciate either confirmation or an explanation of why I might be mistaken.</blockquote><p>Follow me on Medium, and let’s connect on <a href="https://www.linkedin.com/in/dmytro-sirant/"><strong>LinkedIn</strong></a><strong> and </strong><a href="https://x.com/sirantd"><strong>X (Twitter)</strong></a><strong>!</strong></p><img src="https://medium.com/_/stat?event=post.clientViewed&referrerSource=full_rss&postId=2e2bdc06eb02" width="1" height="1" alt="">]]></content:encoded>
        </item>
        <item>
            <title><![CDATA[AWS Solutions Architect — Professional. 2024, Re-certification.]]></title>
            <link>https://medium.com/@sirantd/aws-solutions-architect-professional-2024-re-certification-f7e202ae5d38?source=rss-cd0af9ee77ad------2</link>
            <guid isPermaLink="false">https://medium.com/p/f7e202ae5d38</guid>
            <category><![CDATA[learning]]></category>
            <category><![CDATA[certification]]></category>
            <category><![CDATA[cloud-computing]]></category>
            <category><![CDATA[aws]]></category>
            <dc:creator><![CDATA[Dmytro Sirant]]></dc:creator>
            <pubDate>Fri, 01 Nov 2024 00:36:16 GMT</pubDate>
            <atom:updated>2024-11-01T00:36:16.749Z</atom:updated>
            <content:encoded><![CDATA[<h3><strong>AWS Solutions Architect — Professional. 2024, Re-certification.</strong></h3><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*-C_P7G8XPFyq42ArTIMSlg.png" /></figure><p>Two weeks ago <a href="https://medium.com/@sirantd/my-aws-certification-journey-22030e310287">I completed a three-year journey of becoming 15x AWS Certified</a>. But it’s only a milestone on the way, and I’m not going to stop hardening my skills. This week I re-certified AWS Solutions Architect — Professional and here is my observations.</p><p>In 2021, “Solution Architect — Professional” was my first certification of such a grade. I took the “SAP-P01” version, and now I got the updated “SAP-C02” version.</p><p>This time, I decided not to prepare for the exam for several reasons:</p><ul><li>This certification aligns the most with my technical skills</li><li>I’m doing Well-Architected reviews as part of my work, which require actual knowledge of AWS services</li><li>I was confident in myself</li><li>I was simply lazy 😉</li></ul><p>So, here are my thoughts about the re-certification:</p><ol><li>It’s still in my top three most complex certifications among the “Machine Learning — Speciality” and “DevOps — Professional”.</li><li>Most of the questions are pretty long, as well as answers to pick from. As with any exam, you can easily limit the choice to two questions, but because of the amount of information you need to read and analyse, it’s not as simple 🤯.</li><li>So far, it’s only the certification where my extra 30 minutes as a non-native English speaker was needed. I finished other certifications with 80–90 minutes left, but in this case, I finished the 75th question when the timer showed 26 minutes left, and I had five questions flagged for review.</li><li>Even though this certification is the most relevant to what I do daily, there are still questions about services I’m almost not using, such as CloudFormation and Windows-related services. Sorry, my Microsoft-obsessed colleagues, but I have had a great life since I said goodbye 👋 to Windows in 2006.</li><li>I found that EKS and ECS are now represented equally, and ECS is no longer the default answer to the containerized workload. I’m so happy about it 👏</li><li>Usually, I got results the same day, but this time, it took two days, and I started to worry even though there was no reason for that — I felt confident that I didn’t fail it.</li><li>I did expect that if my first certificate expiration date was 5 Nov 2024 and I re-certified on 30 Oct 2024, the new expiration would be 5 Nov 2027, but it’s not 🤷‍♂️. It is good that I didn’t book my re-certification on an earlier date.</li><li>It was a pleasure to see that I got a higher score than the first certification 🥳.</li></ol><p>So, because of the lower grade certifications, re-certified automatically, I am covered till December 2025, when I’ll need to take care of my “AWS Certified DevOps Engineer — Professional”.</p><p><strong>I still have no decision about the next steps in my learning. Please advise what I should learn next ✍️.️</strong></p><p>Follow me on Medium, and let’s connect on <a href="https://www.linkedin.com/in/dmytro-sirant/"><strong>LinkedIn</strong></a><strong> and </strong><a href="https://x.com/sirantd"><strong>X (Twitter)</strong></a><strong>!</strong></p><img src="https://medium.com/_/stat?event=post.clientViewed&referrerSource=full_rss&postId=f7e202ae5d38" width="1" height="1" alt="">]]></content:encoded>
        </item>
        <item>
            <title><![CDATA[My AWS Certification journey]]></title>
            <link>https://medium.com/@sirantd/my-aws-certification-journey-22030e310287?source=rss-cd0af9ee77ad------2</link>
            <guid isPermaLink="false">https://medium.com/p/22030e310287</guid>
            <category><![CDATA[aws]]></category>
            <category><![CDATA[learning]]></category>
            <category><![CDATA[cloud-computing]]></category>
            <category><![CDATA[careers]]></category>
            <category><![CDATA[certification]]></category>
            <dc:creator><![CDATA[Dmytro Sirant]]></dc:creator>
            <pubDate>Mon, 28 Oct 2024 12:56:56 GMT</pubDate>
            <atom:updated>2024-10-28T12:56:56.520Z</atom:updated>
            <content:encoded><![CDATA[<figure><img alt="" src="https://cdn-images-1.medium.com/max/777/1*WCOUskTBYEWmaf_HrtEdmQ.png" /></figure><p>I’ve just achieved my 15th AWS Certification, and it’s time to reflect on the journey, which has spanned five years. This article aims to give you my <strong><em>subjective</em></strong> feedback on each certificate and its impact on my professional career development so you can decide about your certification journey.</p><h3>Back to the origins</h3><p>I started my IT career back in 1996, and besides the two Brainbench certificates I obtained at the end of 2000, I had no obsession with certifications. Building my own ISP company at the time meant I was enamoured with Cisco and even considered pursuing CCNA and CCNP qualifications — but it remained a pipe dream.</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*SsBfNOjPQScl9TuqeCkdqQ.jpeg" /></figure><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*nxR9t55xVGwbfzAEx7gFlw.jpeg" /></figure><h3>Modern history, how it began</h3><p>Everything changed when I joined <a href="https://opsworks.co/">OpsWorks Co</a> in April 2019. At that point, DevOps was a buzzword, and if you wanted to hire someone with knowledge of Linux, networking, multiple clouds, and conversational English, it would cost an arm and a leg. We realised that without specialisation, we couldn’t hire less experienced specialists and train them quickly enough internally. AWS was the largest cloud provider at the time (a fact that remains true today, five years on), so after some debate, it was chosen as our primary focus.</p><p>If a company providing consulting services in AWS wants confirmation of its expertise, it can become an AWS Consulting Partner. One of the requirements is having a specific number of engineers with AWS certifications, and that’s where my journey began.</p><p>In December 2019, I took the <strong>Cloud Practitioner</strong> free test exam and realised that hands-on experience with AWS, plus common sense, was enough to pass the exam without preparation. On 7th December 2019, I earned my first certificate.</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/300/0*9Bqg2k_fNvF1CYBd" /></figure><p>The following certificate I planned was <strong>Solutions Architect — Associate</strong>, and I decided to complete Neal Davis’s course before taking the exam. It helped organise everything I had learned on the job and filled gaps in services I had never encountered. On 4th April 2020, I earned my second AWS certificate.</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/300/0*jbSIJzlL0L1iRl9D" /></figure><p>Because the entry level for the AWS Partner Select tier was relatively low, my colleague and I only needed to pass the Associate-level exams, so I didn’t push myself to earn more certifications and continued working on various AWS projects for our clients.</p><p>In November 2021, we completed the remaining requirements to achieve the Advanced tier of the AWS Consulting Partner programme, which required at least three engineers with Professional or Specialty certifications. Once again, <a href="https://www.linkedin.com/in/nealkdavis/">Neal Davis</a>’s course helped with preparation, and on 5th November 2021, I achieved one of the most recognised certifications in the field — <strong>AWS Certified Solutions Architect — Professional</strong>.</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/318/0*hhgJLFPz8dmT-ASV" /></figure><h3>Testing the value of certifications</h3><p>With our Advanced partner status secured, we continued hiring new engineers and started evaluating the value of certifications for our hiring process. We found that if a candidate held AWS certifications, we could better assess their knowledge. Instead of gauging their foundational skills, we could start from the level indicated by their certification and save time for other interview questions. Another benefit was that AWS provided virtual credits for each new certificate earned by a partner employee. This enabled us to introduce staff bonuses without additional pressure on the company budget.</p><p>As CTO, I decided to encourage our engineers to pursue AWS certifications by reimbursing certification costs and offering bonuses to those who made the most progress by the end of 2022. To form a personal opinion on the complexity and value of each certification, I also decided to take all of them.</p><p>With <strong>Solutions Architect — Professional</strong> already in hand, I didn’t spend much time studying for the Associate-level exams. I earned <strong>AWS Certified SysOps Administrator — Associate</strong> on 17th October 2022, followed by <strong>AWS Certified Developer — Associate</strong> on 21st November 2022. I then focused on the <strong>AWS Certified DevOps Engineer — Professional</strong>, which I successfully passed on 12th December 2022. The challenging part was learning tools I had never used, such as AWS CodePipeline, AWS CodeBuild, and CloudFormation (I had only used Terraform as an IaC tool).</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/447/1*eLnjkonbnKXXy9LF8of2Uw.png" /></figure><p>Initially, my goal was to learn new things during preparation. However, I adjusted my learning strategy once my goal shifted to earning all certifications to evaluate their complexity and my skills. The more I studied, the more I noticed overlaps between exams. However, I found myself thinking about client cases correlated with exam questions, leading to potential improvements I could implement. Fortunately, I had an extra 30 minutes for each exam as a non-native English speaker, and this side-thinking did not impact the final results.</p><p>Focusing on different KMS types prepared the <strong>AWS Certified Security — Specialty</strong> quickly and straightforwardly, and I earned the certification on 22nd December 2022. Besides the <strong>Cloud Practitioner</strong>, I found this to be the easiest certification, possibly because of the 80% overlap with other AWS exams and the application of basic security principles.</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/318/0*DK5MN1XnmE84sguw" /></figure><p>For <strong>AWS Certified Advanced Networking — Specialty</strong>, I focused on Direct Connect (which I had never used in real life) and experimented with Transit Gateways. I successfully earned the certification on 28th December 2022. While Advanced Networking is often considered one of the most challenging certificates, I didn’t find it difficult, likely due to my previous experience as a Networking Engineer. It’s hard to gauge how challenging networking might be for someone starting their career as a Cloud Engineer without prior experience with BGP on Cisco routers or Zebra.</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/318/0*5Iewh9_ufaN7bLB3" /></figure><p>I expected to pass <strong>AWS Certified Database — Specialty</strong> without much preparation, but I soon realised my knowledge was limited to relational databases and key-value principles. With so many NoSQL, time-series, graph, and ledger databases I had never worked with, it took me about a month to fill the gaps in my knowledge. On 21st December 2023, I earned my ninth certificate.</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/316/0*cDzQ2JtGCDY02Ao-" /></figure><h3>The tiniest overlap with my skills</h3><p>After earning six certificates in five months and doubting I could continue learning at the same pace, I took a break and intermittently watched <a href="https://www.linkedin.com/in/stephanemaarek/">Stephane Maarek</a>’s Data Analytics course. I couldn’t rely on my expertise anymore; only some common sense and understanding of how everything works in IT can help me with further exams. When AWS announced the deprecation of certain certifications in 2023, I made another push toward earning all available certifications. I achieved <strong>AWS Certified Data Analytics — Specialty</strong> on 18th December 2023, <strong>AWS Certified SAP</strong> on 29th January 2024, and the beta <strong>AWS Certified Data Engineer — Associate</strong> on 13th February 2024.</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/444/1*TQvePmhyCs4KrKRBZrwm-w.png" /></figure><p>Preparing for certifications in unfamiliar domains was challenging. Workshops didn’t help much for two reasons:</p><ol><li>I mainly use the AWS web console for billing and Cost Explorer, while everything else is done in code via Terraform or automated with AWS CLI commands.</li><li>Following step-by-step instructions feels tedious to me, offering little creativity and providing results no better than a “Hello World!” application in programming.</li></ol><h3>Terra incognita</h3><p>The final challenge was the AI and ML domain. I couldn’t ignore the missing certifications, as AI had become the buzzword of 2023 and 2024, with almost every startup integrating AI tools. I studied <a href="https://www.linkedin.com/in/stephanemaarek/">Stephane Maarek</a>’s Machine Learning course for six months but struggled to retain all the terminology, abbreviations, and hyperparameters. I made notes and flashcards but still found it overwhelming, much like how some developers might struggle with Advanced Networking. As the beta period for <strong>AWS Certified AI Practitioner</strong> ends, I decided to attempt it and earned the certification on 25th September 2024. With no more excuses for delay, I earned <strong>AWS Certified Machine Learning — Specialty</strong> on 4th October 2024 with a surprisingly good score. While the knowledge was still fresh, I undertook additional training on Bedrock and achieved my final certification, <strong>AWS Certified Machine Learning Engineer</strong>, on 16th October 2024.</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/714/1*6YJZ5Ca4_in_4hy5f9xNLA.png" /></figure><h3>Summary and takeaways</h3><p>This entire journey took five years, though it’s possible to complete in a shorter time with more focused training and less procrastination. If you aim to collect all available certifications, I have good news: as of today, only 12 are in demand. Only two are at the “<strong>Professional”</strong> level and three at the “<strong>Specialty”</strong> level; the rest are either “<strong>Foundational”</strong> or “<strong>Associate”</strong> level. However, the beta <strong>AWS Certified Data Engineer — Associate</strong> and <strong>AWS Certified Machine Learning Engineer</strong> didn’t feel any easier than their “<strong>Specialty”</strong> counterparts. I hope it will change once they move out of beta.</p><h4><strong>Here are some additional takeaways for your journey:</strong></h4><ul><li>Start with the “<strong>Foundational</strong>” certifications, even if your expertise is more advanced. It familiarises you with the exam format, question types, and timing. Moreover, it provides a 50% discount code for your next exam.</li><li>If you are a non-native English speaker — request an extra 30 minutes for your exams. I remember it was useful for me initially, but later, I had about 90 minutes left after revising all the questions.</li><li>Leverage your domain expertise. If you’re a developer, begin with <strong>AWS Certified Cloud Practitioner</strong>, then <strong>AWS Certified Developer — Associate</strong>, and then <strong>AWS Certified Solutions Architect — Associate</strong>. If you’re a Cloud Engineer or DevOps professional, start with <strong>AWS Certified Cloud Practitioner</strong>, <strong>AWS Certified SysOps Administrator — Associate</strong>, and <strong>AWS Certified Solutions Architect — Associate</strong>.</li><li>Be honest with yourself about why you need these certifications. A certificate <strong>without</strong> practical experience is <strong>meaningless!</strong> Even with three AI/ML certifications, I wouldn’t feel confident performing associate-level Machine Learning tasks without supervision.</li><li>Certifications can help stand out during initial CV screening, but the following interview will quickly reveal that you’ve memorised material without a deep understanding.</li><li>Another value of certification is to identify gaps in your knowledge. I’ve come across valuable questions during exams that I researched further afterwards.</li><li>Earning all certifications won’t make you a specialist in every field; you’ll be more of a generalist with foundational knowledge. A focused expert will always outperform you in their niche area.</li></ul><p>Still see value in pursuing certifications? Great! Some efforts from your side and you can achieve it.</p><p>Bon voyage!</p><p>Let’s connect on <a href="https://www.linkedin.com/in/dmytro-sirant/"><strong>LinkedIn</strong></a><strong> and </strong><a href="https://x.com/sirantd"><strong>X (Twitter)</strong></a><strong>!</strong></p><img src="https://medium.com/_/stat?event=post.clientViewed&referrerSource=full_rss&postId=22030e310287" width="1" height="1" alt="">]]></content:encoded>
        </item>
    </channel>
</rss>