Ask HN: AWS cdk, serverless setup advice
I'm doing this part time work with another "senior/staff" developer who got tired of writing lots of code with nodejs and decided the way to go was doubling down on aws - and the various tools provided. His new stack is as follows - dynamodb - nodejs lambda functions - s3 - cdk to setup infra, attach lambda's to various resources nearly all our features rely on the following "common/reuseable" existing code. This is thoroughly tested because they are reused in multiple projects. -> we have a core set of infra/lambda code that does common things like -> authorization/authentication/accounts - using auth0 behind the scenes - for user management -> billing - using a stripe setup. we have ddb table, webhooks and various lambda to support billing based on accounts above -> managing uploads - presigned and direct multipart/form-data uploads and saving metadata in dd -> creating and tracking jobs - via sqs -> have api as a service module that helps us track useage for api's being sold to public For a typical project, we spin up the above existing infra Add custom infrastructure to do things specific to the project This typically involves -> adding an api gateway and attaching lambda handlers -> and so on. My question is Which of you do something similar How do you organize reuseable code - do you create your own custom constructs library? Do you have one giant gateway that custom apis get attached to or you create new gateways to do whatever. 0 comments on Hacker News.
I'm doing this part time work with another "senior/staff" developer who got tired of writing lots of code with nodejs and decided the way to go was doubling down on aws - and the various tools provided. His new stack is as follows - dynamodb - nodejs lambda functions - s3 - cdk to setup infra, attach lambda's to various resources nearly all our features rely on the following "common/reuseable" existing code. This is thoroughly tested because they are reused in multiple projects. -> we have a core set of infra/lambda code that does common things like -> authorization/authentication/accounts - using auth0 behind the scenes - for user management -> billing - using a stripe setup. we have ddb table, webhooks and various lambda to support billing based on accounts above -> managing uploads - presigned and direct multipart/form-data uploads and saving metadata in dd -> creating and tracking jobs - via sqs -> have api as a service module that helps us track useage for api's being sold to public For a typical project, we spin up the above existing infra Add custom infrastructure to do things specific to the project This typically involves -> adding an api gateway and attaching lambda handlers -> and so on. My question is Which of you do something similar How do you organize reuseable code - do you create your own custom constructs library? Do you have one giant gateway that custom apis get attached to or you create new gateways to do whatever.
I'm doing this part time work with another "senior/staff" developer who got tired of writing lots of code with nodejs and decided the way to go was doubling down on aws - and the various tools provided. His new stack is as follows - dynamodb - nodejs lambda functions - s3 - cdk to setup infra, attach lambda's to various resources nearly all our features rely on the following "common/reuseable" existing code. This is thoroughly tested because they are reused in multiple projects. -> we have a core set of infra/lambda code that does common things like -> authorization/authentication/accounts - using auth0 behind the scenes - for user management -> billing - using a stripe setup. we have ddb table, webhooks and various lambda to support billing based on accounts above -> managing uploads - presigned and direct multipart/form-data uploads and saving metadata in dd -> creating and tracking jobs - via sqs -> have api as a service module that helps us track useage for api's being sold to public For a typical project, we spin up the above existing infra Add custom infrastructure to do things specific to the project This typically involves -> adding an api gateway and attaching lambda handlers -> and so on. My question is Which of you do something similar How do you organize reuseable code - do you create your own custom constructs library? Do you have one giant gateway that custom apis get attached to or you create new gateways to do whatever. 0 comments on Hacker News.
I'm doing this part time work with another "senior/staff" developer who got tired of writing lots of code with nodejs and decided the way to go was doubling down on aws - and the various tools provided. His new stack is as follows - dynamodb - nodejs lambda functions - s3 - cdk to setup infra, attach lambda's to various resources nearly all our features rely on the following "common/reuseable" existing code. This is thoroughly tested because they are reused in multiple projects. -> we have a core set of infra/lambda code that does common things like -> authorization/authentication/accounts - using auth0 behind the scenes - for user management -> billing - using a stripe setup. we have ddb table, webhooks and various lambda to support billing based on accounts above -> managing uploads - presigned and direct multipart/form-data uploads and saving metadata in dd -> creating and tracking jobs - via sqs -> have api as a service module that helps us track useage for api's being sold to public For a typical project, we spin up the above existing infra Add custom infrastructure to do things specific to the project This typically involves -> adding an api gateway and attaching lambda handlers -> and so on. My question is Which of you do something similar How do you organize reuseable code - do you create your own custom constructs library? Do you have one giant gateway that custom apis get attached to or you create new gateways to do whatever.
Hacker News story: Ask HN: AWS cdk, serverless setup advice
![Hacker News story: Ask HN: AWS cdk, serverless setup advice]()
Reviewed by Tha Kur
on
June 14, 2025
Rating:
No comments: