Yahoo

The gym is open all day, the cafeteria food is better than average (and apparently free now), and the coffee is free.

Engineering is a mess. Officially supported software was super outdated (Apache 1.3? FreeBSD 4.x?). Teams frequently duplicated effort. Official standards (No more FreeBSD!) conflicted with reality (Work in mail? Write your code on FreeBSD). Want to get a small change pushed out live? Hope you like meetings and overhead.

The team I was on was "Agile," which meant we sat through 4 hours a week of sprint-related meetings.

I was hired to rebuild an existing system only to find out after I started that I wasn't allowed to change said system because too many people depended on it being exactly the way it was. (Said system came originally from another company Yahoo acquired in the late 90's - think DHTML and Java applets for scrolling text on the screen and you're in the ballpark).

So after two months, I had nothing to do other than use the gym and hunt for jobs.

You should seriously nuke most engineering projects from orbit. This isn't an attempt at sarcasm or humor; you'd have an easier time rebuilding entire teams, products, and organizations from scratch than you will trying to improve things by changing a piece here or there. Think hard about why you're buying up all of the startups that you are. How have <10 person teams managed to come up with technology that blows away anything your product team has come up with in the last decade? It's because they're not carrying around all of your luggage.