Business process underpin critical enterprise operations across virtually all industries including health care, finance, and retail, and optimizing and automating these processes can reduce cost or improve customer experience. However, the most common automation approaches, such as structured workflow management systems, operational rules, and robotic process automation (RPA), require skilled developers to implement and only tend to only address the common, anticipated cases. In reality business processes exhibit a long tail of rare or unique situations that often fall on human knowledge workers to handle.This talk will present an approach to business automation that allows business users to use a conversational interface to interact with, compose, and customize automations. This solution relies on recent advances in AI, such as natural language understanding and explainable AI, to enable non-programmers to more effectively and efficiently manage the long tail of situations that are difficult to automate. This talk will also highlight a number of research challenges to fully realize this solution.

I am a research scientist at IBM Research AI in the IBM T.J. Watson Reseach Center. My research interests are broadly in middleware systems, including tools and infrastructure to support a variety of applications. In particular:

• Platforms for training and serving large machine learning models, as well as applications of deep learning, and bringing more attention to these challenges to the middleware and systems communities.
• Business process analytics, including process mining, clustering, and exploiting blockchain technologies.

I completed my PhD and MASc degrees in Computer Engineering at the University of Toronto, and my BASc degree at the University of Waterloo.

You can find a list of publications at my Google Scholar and DBLP pages.