Central to the development of long-distance quantum communications is the concept of quantum repeater. It consists in dividing a long communication channel into various shorter segments over which entanglement can be faithfully distributed. Adjacent segments are then connected by entanglement swapping operations. To be scalable, this approach requires quantum memories, which enable quantum states to be stored at each intermediate node. In this context, the LKB team has been developing non-degenerate sources of entangled photon pairs compatible with both telecom networks, and an atomic quantum memory. This quantum memory based on a cold atomic ensemble in the group enables qubit storage with an overall efficiency close to 90% mark for entanglement storage between two memories. The work is now focusing on two directions. A first one is to improve the figures of merit, including the efficiency, and to interface it with an atomic quantum memory. A second one is the implementation of quantum networking protocols, from photonic teleportation on a dedicated fiber network on the Jussieu campus, to the demonstration of a 50-km telecom quantum repeater link relying on two distant quantum memories and frequency non-degenerate photon pair sources. Part of the work will be led in collaboration with the startup company Welinq.