The Smart City paradigm is progressively shaping the way we inter-act with other citizens, with institutions, as well as the way in which all kind of resources are managed in an urban setting. However, the high cost of dedicated sensing, computing and communication infrastructure, represents one of the main obstacles to the adoption of the Smart City paradigm in small cities and large, dis-tributed districts, where resides the vast majority of the world population. In this paper we present a first implementation of a platform for nomadic sensing, which exploits the moving infrastructure of a city in order to implement the sensing, computing and communication functionalities required by smart city services, in a cost-effective way. This paper describes the architecture of the NOSE system, and the main lessons learned during the first steps of its implementation, deploy-ment, and experimental validation. In particular, we show that the main technical challenges are due to the lack of devices conceived for reliable mobile operation (sensing, computing) and to the difficulty in achieving high levels of accuracy in measurements while moving.