Hello, I've been working in a big pharmaceutical company in France where I introduced NAMD. I recently left and created my own company, where we adapted NAMD (and to be more precise, the usage it has of Charm++ parallel objects) to be used for financial forecasts. Our infrastructure is currently rather well scaled for our needs : more or less 50 CPU Linux cluster. We are now envisaging to work in sports events forecasts. You will find more information on Sentimancho's website