Testing wireless networks is a challenging task, so many research papers limit their evaluation section to simulation. At the same time, several large-scale publicly shared testbeds such as ORBIT, w-iLab.t, or IoT-LAB propose valuable means for validating new protocols. They make experimenting possible and support some level of experiment reproducibility. However, they offer operating conditions that are fixed and significantly differ from real-world deployments. As the performance of wireless networks strongly depends on the platform, topology, and operating conditions, validating protocols in artificial conditions may not lead to meaningful results. In this paper, we describe WalT, a reproducible platform to run reproducible experiments. WalT nodes are single-board computers on which users can deploy their OS (filesystem, kernel) packaged as a docker image for easy customization and sharing. With low-cost small-sized standard components and free software, researchers can easily reproduce their own WalT platform to validate results in real-world conditions. WalT can support mobile demos that you can bring around in your backpack. The total control on WalT nodes allows setting up diverse experiment scenarios ranging, for instance, from Wi-Fi handover measurements to evaluating routing protocols in wireless sensor networks.