Six flexible force sensors, two on the backrest and four on the seat, were embedded in the upholstery of an off-the-shelf office chair to enable non-intrusive monitoring of sitting postures. Besides the sensors, the monitoring platform comprises an Arduino Nano microcontroller with Wi-Fi transmitter, embedded on the chair, a Wi-Fi receiver communicating with a remote server and a Graphical User Interface (GUI) showing real-time readings. Approximately 26,000 observations corresponding to 9 different postures were collected, labelled and classified using supervised machine learning. The results show that only a subset of the 6 sensors is needed for predicting these 9 sitting postures with high accuracy. This opens up the possibility for intelligent, real-time monitoring systems that can improve safety and wellbeing of today's office workers.