
First I find a task that overlaps with new knowledge.
Examples:
- Query optimization - improving slow query
- Clean code/architecture - completely new feature
- TDD - new feature, legacy code
- Refactoring - a feature that could reuse some code

Don't jump straight into it.
Tell them:
Why would I do it?
How would I do it?
Ask them:
Do they have any better ideas?
Do they see any drawbacks?
Do they understand what I want to do?

After your team agreed to try something use your new knowledge to solve the problem.
In MR/PR describe what you did and why to let the reviewer(s) know about your intentions.

After a new code is in production wait some time to see what's going on.
Check:
Can you easier develop things related to your change?
Are there fewer/more bugs than usual?
Is there any unexpected consequence?

When your tracking questions are answered evaluate the approach with your teammates.
- it's an improvement - start using it
- there's no change - track it a bit longer
- it's not an improvement - don't use it again