My name is Ilya Valmianski and I am a machine learning scientist at Curai Health. My background is in physics, for which I recieved a PhD from UC San Diego in 2017. However, my true passion was in understanding and modeling data. So, after graduation I radically changed my career direction and joined the Medical Informatics team at Kaiser Permanente, where I worked on medical NLP and clinical decision support. The culmination of my work there was development of SmartTriage, which is a dynamic pre-visit questionaire trained on tens of millions of medical progress notes. On the EHR side, SmartTriage pre-writes progress notes for physicians as well as provides clinical decision support. A highly successful project, it was deployed widely in KP Southern California region, which at the time makes it one of the largest patient facing machine-learning-based applications. After Kaiser Permanente, I joined Curai Health, where I am now. At Curai I work on developing NLU and NLG models for supporting medical chat with patients as well providing clerical and decision support to physicians. My main interest recently has been on utilizing external knowledge with deep learning models. This includes knowledge driven pseudo labeling and active learning, supporting NLU models with external ontologies, and constraining generation by injection knowledge either as training regularization or as generation constraints.
Find me on Twitter, LinkedIn, Google Scholar, or Github. My resume can be found here.