Sensors networks are the backbone of large sensing infras-tructures such as Smart Cities or Smart Buildings. Classical approaches suffer from several limitations hampering developers' work (e.g., lack of sensor sharing, lack of dynamicity in data collection policies, need to dig inside big data sets, absence of reuse between implementation platforms). This paper presents a tooled approach that tackles these issues. It couples (i) an abstract model of developers' requirements in a given infrastructure to (ii) timed automata and code generation techniques, to support the efficient deployment of reusable data collection policies on different infrastructures. The approach has been validated on several real-world scenarios and is currently experimented on an academic campus.