June 2018 New paper on backward reachable curricula for robotic reinforcement learning uploaded to arXiv, with code!
April 2018 I will be an Aeronautics and Astronautics PhD student at Stanford this fall!

About Me

I obtained my Master's in Computer Science from Stanford University in 2018, specializing in Artificial Intelligence (AI).

Prior to that, I obtained my Bachelor of Applied Science with High Honours in 2016 from the University of Toronto's rigorous Engineering Science program, majoring in Electrical and Computer Engineering with a Robotics/Mechatronics minor.

Resume/CV (updated June 18, 2018)


  • Commencing September 2018
    Stanford University
    Aeronautics and Astronautics - Doctor of Philosophy (PhD)
    Stanford, CA - USA
    • Conducting research at the intersection of robotics and deep learning under Prof. Marco Pavone.
  • September 2016 - June 2018
    Stanford University
    Computer Science - Master of Science (MS)
    Stanford, CA - USA
    • Conducted research in machine learning, computer vision, robotics, and data science.
    • I have had the pleasure of working in the following labs:
  • September 2012 - June 2016
    University of Toronto
    Engineering Science - Bachelor of Applied Science (BASc) with High Honours
    Toronto, ON - Canada
  • September 2003 - June 2012
    Bayview Glen
    Ontario Secondary School Diploma (OSSD)
    Toronto, ON - Canada
    • Top of Grade (highest overall average) for Grades 11 and 12.
    • Graduated as an Ontario Scholar and Governor General’s Academic Medal holder.
    • Graduated as an Advanced Placement Scholar with Honour.


  • BaRC: Backward Reachable Curriculum for Robotic Reinforcement Learning
    B. Ivanovic
    , J. Harrison, A. Sharma, M. Chen, M. Pavone
    [BibTex] [PDF] [arXiv] [code] arXiv 2018
  • Generative Modeling of Multimodal Multi-Human Behavior
    B. Ivanovic
    , E. Schmerling, K. Leung, M. Pavone
    [BibTex] [PDF] [arXiv] [code] arXiv 2018
International Peer-Reviewed Conference Proceedings
  • ADAPT: Zero-Shot Adaptive Policy Transfer for Stochastic Dynamical Systems
    J. Harrison, A. Garg,
    B. Ivanovic
    , Y. Zhu, S. Savarese, L. Fei-Fei, M. Pavone
    [BibTex] [PDF] [arXiv] Int'l Symposium on Robotics Research (ISRR) 2017
Blog Posts

Work and Research Experience

  • June 2017 - September 2017
    Prime Air SDE Intern
    Amazon.com - Seattle, WA - USA
    • Worked with Principal Research Scientist Ishay Kamon in the Autonomy team.
    • Designed and implemented a novel state-of-the-art deep learning approach for a specific computer vision task within the team, outperforming existing models by 10x.
    • The project was completed successfully and a full-time Research Scientist return offer was extended.
  • April 2017 - June 2017
    CS231A Course Assistant
    Stanford University - Stanford, CA - USA
  • January 2017 - June 2017
    Independent Research Project
    Stanford University - Stanford, CA - USA
    • Worked in the Computer Vision and Geometry Lab (CVGL) and Autonomous Systems Lab (ASL) with Professors Silvio Savarese and Marco Pavone on making reinforcement learning more robust with control theory.
    • Tackled the problem of policy transfer in reinforcement learning, applying model-predictive control to provide safety guarantees when transferring a learned policy from one environment to another with different dynamics.
    • Our work was accepted to the International Symposium on Robotics Research (ISRR) 2017, held in Puerto Varas, Chile.
  • September 2016 - June 2017
    Research Assistant
    Stanford University - Stanford, CA - USA
    • Worked in the Stanford Network Analysis Project (SNAP) Lab with Professor Jure Leskovec on analyzing large-scale physical activity data with modern data science methods.
    • Efficiently cleaned, preprocessed, and distilled 3 TB of user physical activity data from over 2 million users of a mobile fitness app. Obtained scientific results with data visualization, statistical analyses (such as regressions and confidence metrics), and computational methods (including hierarchical bootstrapping).
    • Showed significant results relating a location’s walkability, weather, and climate to an individual’s physical activity. This work has very wide implications, as physical inactivity is a major global pandemic responsible for over 5 million deaths per year.
  • September 2016 - December 2016
    Independent Research Project
    Stanford University - Stanford, CA - USA
    • Worked with Professor Andrew Ng on neural language correction, a method of correcting grammar and spelling mistakes using deep sequence models.
    • Augmented state-of-the-art deep neural language models with subword units and experimentally verified resulting model performance.
  • May 2016 - August 2016
    Prime Air SDE Intern
    Amazon.com - Seattle, WA - USA
    • Worked with former NASA Astronaut Neil Woodward in the Flight Test team.
    • Designed and built fault-tolerant, scalable software and hardware to autonomously collect and process relevant flight test data from numerous locations for internal consumption.
    • The project was completed successfully and a return offer was extended.
  • January 2016 - May 2016
    CSC411 Teaching Assistant
    University of Toronto - Toronto, ON - Canada
  • September 2015 - May 2016
    Undergraduate Thesis
    University of Toronto - Toronto, ON - Canada

    Worked with Professors Raquel Urtasun and Sanja Fidler to:

    Create a semantically labelled 3D map of an outdoor environment from video recorded by a quadrocopter and augment that map with cartographic data (ie. from OpenStreetMap).

    More specifically, the goals of the project were:

    • Fly a quadcopter around an outdoor environment and, from video recorded during the flight, create a 3D reconstruction of the environment. The LSD-SLAM algorithm was used for this since the cameras available to us were monocular.
    • Design and implement a data annotation framework to support and automate the labeling of collected data.
    • Label objects within the 3D representation of the environment, such as people, trees, buildings, etc... For this I formulated a CRF, incorporating cartographic data.

    Project results and reports can be viewed at this Google Drive link.

  • May 2015 - August 2015
    Summer Research Intern
    ETH Zurich - Zurich - Switzerland
    • Worked with Professor Raffaello D’Andrea in the Institute for Dynamic Systems and Control, specifically the Flying Machine Arena.
    • Removed superfluous code from an open source motor controller and implemented new features such as motor calibration, emergency safety states, and a better motor startup routine in C.
    • Simulated dynamic motor and propeller system responses in Python.
    • Technology used: STM32 C Code, Motor Controller PCB Chips, Quadrotor Flying Vehicles.
  • May 2014 - July 2014
    SDE Intern
    Amazon.com - Seattle, WA - USA
    • Worked in the Demand Forecasting team creating a real-time simulation tool. The project was completed successfully and a return offer was extended.
    • Worked with Big Data, using the Hadoop framework (MapReduce, HDFS, etc.) to process large amounts of simulation data generated by a machine learning module.
    • Created a web service with Spring, designed and implemented a website UI with GWT, and used the AWS SDK to store and retrieve data from S3.
    • Gave a presentation to 100+ Amazon employees regarding my project and its design, implementation, and performance.
    • Concepts employed: Big Data, Highly Scalable Distributed Systems, and Data Mining.
  • May 2013 - August 2013
    Agile Engineer Co-op
    Xtreme Labs (now Pivotal) - Toronto, ON - Canada
    • Learned Agile development methodologies and applied them to real-world projects.
    • Provided input on and developed an app that has 1,000,000+ installs.
    • Technology used: Android SDK for apps and HTML/CSS3/PHP for website development.


  • May 2016
    Engineering Science Award of Excellence
    University of Toronto - Toronto, ON - Canada

    Received for maintaining a CGPA greater than 3.90.

  • May 2016
    Computer Science TA Award
    University of Toronto - Toronto, ON - Canada

    Received for being the best Computer Science TA in the Winter 2016 semester.

  • April 2016
    NSERC Master's Postgraduate Scholarship (CGS-M) (Declined)
    National Sciences and Engineering Research Council (NSERC) - Canada

    The CGS-M Program provides financial support to high-calibre scholars who are engaged in eligible master’s programs in Canada (more information here).

  • September 2012 - June 2016
    Dean’s Honour List
    University of Toronto - Toronto, ON - Canada

    Placed on the Dean’s Honour List for all undergraduate semesters.

  • September 2012
    University of Toronto Scholarship
    University of Toronto - Toronto, ON - Canada

    Received for being one of the top 300 entrants to the University of Toronto in 2012.