Modern cities are facing tremendous amount of information, captured from internal in- frastructures and/or exogenous sensors, humanincluded. This talk presents how big and het- erogenous city data has been captured, represented, unified to serve one of the most pressing city objective: improving quality of city, in particular how understanding and reducing traf- fic congestion. We will also present lessons learnt from the deployment of our system and experimentation in Dublin (Ireland), Bologna (Italy), Miami (USA) and Rio (Brazil).