In our days there are a few occasions when you wrote all the code. I will give 2 examples:
1. Sorting an array
In the late 90 I will do a Quick sort implementation . In our days, the .NET the Array Sort https://docs.microsoft.com/en-us/dotnet/api/system.array.sort?view=net-5.0 has many overloads – pick your poison
2. Make a persistent searchable chat between users of the application
In the 90 I will do an implementation with some network file and/or opening a communication channel ( or, if I were picky , with a database) . Now I have so many options to do the job straight away: databases( SqlServer , Mongo ), communications ( RabbitMQ , SignalR, GRPC ) , languages ( .NET Core for me ), architecture ( communicating from one client directly via REST / GRPC / WCF / ) , testing ( MSTEST, XUnit, NUnit…) , compiling and deploying ( GitHub Actions, Azure Devops …). And I do not want to comment about the vast amount of components on Nuget / Npm .
So in our days counts not only knowing a language and how to do algorithms , but also knowing a large amount of components that can make your work easy to implement.
Note: For sure, there are also programmers that works to the Google/Bing Search Engine. That are working to innovation at ML / AI . But I think that the most majority of programmers work for some kind of business – and they are much more productive if they are not a DIY/NIH programmers )