In the public sector, big data holds many promises for improving policy outcomes in terms of service delivery and decision-making and is starting to gain increased attention by governments. Cities are collecting large amounts of data from traditional sources such as registries and surveys and from non-traditional sources such as the Internet of Things, and are considered an important field of experimentation to generate public value with big data. The establishment of a city data infrastructure can drive such a development. This paper describes two key challenges for such an infrastructure: platform federation and data quality, and how these challenges are addressed in the ongoing research project CPaaS.io.