I think a great architect knows when to use a battle-tested tool and when to roll-out his own. There is no shame in using something that has a great track record of not failing you. Building a blueprint is not about writing a new formula for a brick, but rather using the best brick on the market and focusing on making the building as usable and efficient as possible.
Honestly, I disagree with your statement. The tsunami of FOSS is sometimes an advantage as well, because people can easily migrate off of legacy architecture that they were hired to fix in the first place and pick one to our liking. Since I do not know your niche or how much experience you have, it's hard for me to guess whether you have worked as a contractor somewhere, but...
I had a guy re-inventing his own Celery with a single worker in the same thread as main process. Of course it failed spectacularly :) I then also had an example of someone deliberately not using already battle-tested auth libraries for social auth but creating his own and of course he managed to expose things that were not needed to the client app and the company got hacked. He spent a month building it, instead of just plugging in an already built library that is used in production by millions of people.
Another developer yet built his own framework which was never stable and costed the company tens of thousands of dollars for the three months it was in production (plus three for development).
These are not some small start-ups or dev-shops, but rather serious companies that had devs that were alumni of Big 4. I think I am not the only one with this experience.
For me that is heresy - similar to not using a great standard library in a any language, but rather building your own libs/wrappers. Now, I have nothing against modifying a library or optimizing it for your use case, however starting things from scratch is in my opinion a huge cost, unless your company has time and money to throw.