Panda's et all do investigate Data. There are tools in Ruby for Datascience but the resources are thin or completely lacking.
A scraper using Scrapy. Had a tool in Ruby with Hydra (multitreaded) and message cueus and whatnot to speed it up. Spidy is just faster. Not because of the language, but because the tooling is much further along and has far more effort poored into.
Server/infra management: ported from chef to ansible. Not for speed, but because stability and availability (of a.o. resources). Unexpected side-effect was that ansible was actually faster and used far less resources on the servers (chef-deamon was notoriously bad). Not sure why, but I don't think its the language design.
FluentD, moved to flowgger and plain-old-syslog. FluentD is an awesome piece of software, but it was hogging our (micro)servers enough to be noticable, measurable and show up in the billing (as higher CPU/memory vpses).