[redis_mlc]

I did some work for the Recommendation team (the oversized checks were still arrayed along the wall when I was there.)

The later winning entries were too compute-intensive to implement, or not enough of an advantage over the existing engine to justify more compute.

So I would say it was a real competition, not just PR, even though a particular solution wasn't used.

(Factoid: It was the the last team to start the migration to Cassandra.)