About

Joel Kaardal is a data scientist and machine learning engineer with a passion for innovation. With over 5 years of relevant experience, he has worked on a range of projects that leverage data and machine learning to build smarter products and uncover novel insights. Joel's expertise lies in machine learning and data science, but he is also a capable software engineer with the flexibility to bring research to life by writing production code. Currently, Joel works as a Machine Learning Research Engineer at doc.ai, a startup where he has worked on several projects, including a mental health chatbot, an ad tech platform, and a medication aide. During his time at doc.ai, Joel has designed a probabilistic consumer/campaign matching algorithm that maximizes revenue for an ad tech platform. He has also conceptualized an artificial memory system to extract interpersonal relationship data from natural language for recall in a cognitive-behavioral therapy chatbot session. Joel has led a project to build emotion and sentiment analysis models that exceeded the performance of available APIs in the mental health domain. Additionally, he has built a content-based recommendation system using natural language data annotated via Mechanical Turk. Joel has also prototyped a computer vision/NLP app to extract metadata from prescription bottles via mobile phone cameras, and the model is currently deployed on the Sharecare mobile platform. Prior to his current role, Joel worked at Level, where he was a Data Scientist/Software Engineer. During his time at Level, Joel designed the LEVEL oven’s image classification system that infers the suggested identity of over 100 different foods with >95% precision and leverages customer feedback to continuously improve. He also conceptualized and implemented an experimental apparatus to generate pixel-precision ground truth depth data to train a deep stereo vision model in TensorFlow. Joel wrote software to optimize and evaluate food detection and semantic segmentation models that locate food to pixel precision within the oven. He managed crowdsourced annotation of detection and segmentation data, reducing the costs of building a quality dataset. Additionally, Joel developed the first production version of the LEVEL oven’s backend software in python. Joel completed his Doctor of Philosophy - PhD in Physics from the University of California, San Diego, in 2017. He also holds a Master of Science - MS in Physics from the same university. Joel earned his Bachelor of Science - BS in Physics and Chemistry from the University of Minnesota in 2010. Joel's tech stack includes ML, Research Scientist, Data Scientist, Mobile, Sentiment Analysis, Recommendation system,