During my PhD, my collaborators and I trained the first successful deep acoustic models for automatic speech recognition. I also led the team that won the Merck molecular activity challenge on Kaggle.