I'm a professional C#/.NET engineer with more than a decade of development experience. I've used AWS storage services through the SDK in previous projects, but the infrastructure itself was usually handled by DevOps or other team members. Now I want to become comfortable with AWS infrastructure as code using CDK, especially for deploying databases, queues, Lambda functions, API Gateway, logging, and related services.
I've already set up CDK and deployed a TypeScript tutorial stack. I also created a stack for a C# Lambda packaged locally without Docker, connected it to API Gateway, and configured log groups with a one-day retention period. Much of my learning has come from debugging deployments, examining CloudFormation output, and using synthesis and verbose deployment commands. I'm currently learning about CloudFormation, VPCs, networking, IAM permissions, and policies, and I'm working through CDK examples and AWS Builder resources.
For developers coming from a software engineering background, what learning path, projects, or resources would you recommend for becoming productive with CDK and AWS infrastructure? Are there concepts or approaches I should prioritize or avoid?
4 Answers
A key CDK concept is that your program generally runs during synthesis; it doesn’t directly create resources when you write expressions such as `new Bucket()`. CDK turns your code into a CloudFormation template, and CloudFormation performs the deployment later. Because of that, values such as a database endpoint may not exist during synthesis, and ordinary logging won’t show the deployed value. Use CDK tokens and supported helper functions when passing resource attributes between constructs, and inspect the synthesized template to understand what will actually be deployed.
Also spend time on the AWS fundamentals behind CDK: IAM, networking, security groups, availability, backups, encryption, quotas, and service limits. CDK becomes much easier once the underlying cloud architecture is familiar.
For safe experimentation, consider running an AWS-compatible local emulator such as LocalStack. It lets you practice parts of the infrastructure workflow without constantly creating real resources or incurring charges. It won’t reproduce every AWS behavior perfectly, so you’ll still need to validate important designs in an actual AWS account.
Don’t feel obligated to switch to Terraform just to learn infrastructure as code. Terraform is useful and widely used, but CDK is a reasonable choice if you already know TypeScript and your target is AWS. The main advantage is that you can create readable abstractions with a familiar programming language while still producing CloudFormation resources. Learn CloudFormation’s model and AWS architecture alongside CDK, rather than treating CDK as a replacement for that knowledge.
The most effective approach is usually to build something real and learn as you hit problems. Tutorials are useful for getting started, but they rarely cover the details you need in an actual project. Pick a small application, deploy it, inspect the generated resources, and troubleshoot each issue as it appears. Repeating that process builds familiarity much faster than watching several hours of videos.
That synthesis-versus-deployment distinction is becoming clearer to me. I’m treating CDK as a declarative layer over CloudFormation, even though the code is written in TypeScript. It still catches me occasionally, but inspecting the synthesized template has helped.