Common Lisp, surprisingly, does not have that many applications in most Data Science, not only from what I do; but the entire Data Science community, as well.
Why is that? Well I cannot speak for everyone, but in my experience the biggest problems with Lisp are as follows:
Data structures in Lisp are generally just lists, which is both good and bad. It is great for something with repetitive elements, not so great for a constructed type with fields.
Dependencies in Lisp are an absolute nightmare, usually they are managed through QL ( QuickLisp ), which is certainly a cool system, but also requires a lot more hassle -- and does not have source-able environments.
All in all, I have asked this same question you are asking me before, but I think it really comes down to the package system of Lisp being inferior to that of Julia's. Hope that made some sense.