I think you misunderstand what memory stores like Redis or memcached can do for you.
They are not connected to PostgreSQL or any other RDBMS. It is the job of your application to write the data in both stores: the permanent store (PostgreSQL in your case) and the transient one (Redis).
Redis and memcached do not offer many connectivity features. For caching, they have a rather simple and unobtrusive behavior. The smart part to manage the cache is on application side, and it is your job to define it.
You can imagine several strategies:
each time you write in the RDBMS, you write in the cache. Each time you need to read the data, read first in the cache. If nothing is found, read from the RDBMS and write it back to the cache.
each time you write in the RDBMS, you just invalidate the data in the cache (delete it). Each time you need to read the data, read first in the cache. If nothing is found, read from the RDBMS and write it back to the cache.
The tricky part is to clearly define a policy regarding the consistency of the data (between the RDBMS and the cache).