TrainAR is an augmented reality feedback platform for providing real-time feedback to novice healthcare practitioners during training in specialized clinical procedures. Novice practitioners frequently acquire the motor skills necessary for these specialized clinical procedures via simulation training. Simulation pedagogy requires that the learner receive immediate feedback and post-training debriefing to best internalize the learning objectives. Instructors and task trainers are often utilized in tandem to allow deliberate practice for novices in specialized clinical procedures. These methods, however, are limited by scarce institutional resources and system limitations to adaptation. In this work we explore the possibility of augmenting clinical procedure training with real-time feedback using augmented reality (AR), machine learning, and computer vision technologies.