What: Lots of small data analysis projects, some large web projects. All projects are in git repos or in our division's SVN. My work tends to involve ETL data flows, statistical analysis, web development and data visualisation. I'm not "full stack" as I never really play with hardware or do much systems administration. I do get to control a nice big, enterprisey Oracle data warehouse, with a few hundred users, but I've hardly played with nosql, redis, mapreduce, etc.
Most of my problems can be solved in-memory, and most of my data fits on a single disk.
How: SQL. SQL. More SQL. PL/SQL. Python (Pandas, scipy, numpy). R (these days only when I can't find a python lib). Javascript (just jquery, underscore and D3). Windows 7 while at work, Ubuntu + i3 wm at home. Edit in vim with very few plugins, everything else is done in zsh/powershell. IIS for production, Pyramid/Pylons Project for quick prototyping. A wide variety of RDBMS, but chiefly Oracle. I do lots of my templating/css in-browser using developer tools (FF, Chrome, IE9).
My typical web stack is static html and javascript, with json served up out of purpose-specific database views. No need for an MV* framework, since I'm almost never dealing with transactional data.
I also use Excel for a) doing quick data transforms b) building interactive, distributable data analysis apps in less than an hour. I'm trying to build an excel replacement with vim bindings and LISPy worksheet formulas.