I used Clojure a lot back in the beginning (contributed a little money, used Clojure fairly intensely on two long term consulting projects). Before that I used Rich's Common Lisp to Java bridge project. I think his impact goes far beyond just Clojure users, having helped push immutable by default, etc. as standard good practices.
I moved away from Clojure because I wanted to move away from the JVM. I did a fun project, a nutrition/cooking web app [1] about 12 years ago in Clojure, but I am slowly porting that from Clojure to Common Lisp for another example program for my book [2]. Perhaps not as well known, but Clojure joins other less commonly used languages like Haskell as languages with fairly good support for deep learning. I feel like any language that does not have a good deep learning story will be limited. Also interesting that Swift has such good access to Apple's deep learning libraries and also a version of TensorFlow. I encourage Clojure developers who work on deep learning support!
You can subscribe to Dragan Djuric's deep learning book [1].
He is using the subscriptions to fund work on Deep Diamond [2], a fast Deep Learning framework with GPU support for Clojure.
I moved away from Clojure because I wanted to move away from the JVM. I did a fun project, a nutrition/cooking web app [1] about 12 years ago in Clojure, but I am slowly porting that from Clojure to Common Lisp for another example program for my book [2]. Perhaps not as well known, but Clojure joins other less commonly used languages like Haskell as languages with fairly good support for deep learning. I feel like any language that does not have a good deep learning story will be limited. Also interesting that Swift has such good access to Apple's deep learning libraries and also a version of TensorFlow. I encourage Clojure developers who work on deep learning support!
[1] http://cookingspace.com [2] https://leanpub.com/lovinglisp