In-ear wearables, called Hearables, have been introduced for the monitoring of several physiological signals, \textit{inter alia} ECG. However, in-ear ECG has a smaller amplitude and lower signal-to-noise ratio than standard ECG, which makes it difficult to automatically detect R peaks with standard algorithms. The solution for these problems may be the use of specially designed R-peak detectors, such as a Deep matched filter (DMF). The DMF is a deep neural network model designed to search for matches with an ECG template pattern in the input signal. The DMF operates as a neural network Matched Filter, trained to analyse in-ear data. In this study, we deploy the DMF in mobile aplication for online R-peak detection from in-ear ECG. To our knowledge, this is the first online R peak detector dedicated to analysis of in-ear ECG. We transfer the DMF into the Android mobile application, that communicates with the BioBoard, a multi-sensors hearable platform developed in our lab. We conduct a computation time estimation for the implementation and analysis of DMF vulnerability to various types and levels of artefacts that can disturb ECG signal. We found that the baseline wander artifacts slightly impact the model's performance. Motion electrode and muscle artifices disturbed the model's sensitivity and precision when signal to noise ratio drop bellow 6 dB. The average time of the single feed-forward model run, analysing 2 seconds of an ECG, on Samsung Galaxy Tab S8 was 7.46 ms. The model can be used for online detection of R peaks in-ear ECG.