Skip to content

About Me

My name is Maxime Chevalier-Boisvert. I completed a PhD degree with a concentration in compiler design at the Université de Montréal. My thesis was focused on Just-In-Time (JIT) machine code optimization techniques for dynamic programming languages (e.g.: JavaScript, Python, etc.). I’m also interested in programming language design, machine learning, 3D rendering, game development, music, sound synthesis, microcontrollers and electronics. This blog is an outlet for me to talk about my research, technological interests, hobbies and ideas. Most of my recent programming work is open source and available on GitHub.

Publications

Combating False Negatives in Adversarial Imitation Learning. Deep Reinforcement Learning Workshop at NeurIPS 2020.

DeepDrummer: Generating Drum Loops using Deep Learning and a Human in the Loop. Joint Conference on AI Music Creativity 2020.

Options of Interest: Temporal Abstraction with Interest Function. Association for the Advancement of Artificial Intelligence (AAAI) 2020.

BabyAI: First Steps Towards Grounded Language Learning With a Human In the Loop. International Conference on Learning Representations (ICLR) 2019.

The AI Driving Olympics at NeurIPS 2018. RSS Workshop on New Benchmarks, Metrics, and Competitions for Robotic Learning 2018.

Interprocedural Type Specialization of JavaScript Programs Without Type Analysis. European Conference on Object-Oriented Programming 2016.

On the Fly Type Specialization without Type Analysis. Chevalier-Boisvert, M.  PhD thesis. Université de Montréal. December 2015.

Simple and Effective Type Check Removal through Lazy Basic Block Versioning. European Conference on Object-Oriented Programming 2015.

Bootstrapping a self-hosted research virtual machine for JavaScript: an experience report. Dynamic Language Symposium 2011.

Optimizing MATLAB through just-in-time specialization. Compiler Construction 2010.

McVM: an optimizing virtual machine for the MATLAB programming language. Chevalier-Boisvert, M. Master’s thesis. McGill University. August 2009.

Talks

ZetaVM, a Platform for Programming Language Innovation. PolyConf. July 2017 (slides, video).

Interprocedural Type Specialization of JavaScript Programs Without Type Analysis. ECOOP. July 2016 (slides, video).

Simple and Effective Type Check Removal through Lazy Basic Block Versioning. ECOOP. July 2015 (slides, video, poster).

Lazy, Incremental JIT Compilation with Basic Block Versioning. Northeastern University. December 2014 (slides).

Refactoring JIT Compilation. DConf. May 2014 (slidesvideo).

Dynamic Languages need Dynamic Compilers. mloc.js. February 2014 (slides, video).

Fast and Dynamic. Strange Loop. September 2013 (slides, video).

Higgs, an Experimental JIT Compiler written in D. DConf. May 2013 (slides, video).

Higgs, a Monitoring JIT for JavaScript. Air Mozilla. February 2013 (slides, video).

21 Comments
  1. John Wait permalink

    since you work on compilers, thought you might find this article by Rob Pike to be of interest: http://www.informit.com/articles/article.aspx?p=1941206

    “…I was pair programming with Ken Thompson on an on-the-fly compiler for..”

  2. What features in Ruby/Python/Javascript are particularly bad for type inference?

    • Dynamic loading of code, unrestricted eval, dynamic addition and removal of object fields. The JavaScript “undefined” type and the lack of exceptions thrown on errors.

  3. Congrats on the JavaScript compiler work!

  4. Dear Maxime, i am working on a new deductive computer language that outputs to Adobe AiR and Javascript, so that the graphical interactive programs you write can run on 5 different platforms: Mac, Win, iOS, ANdroid, and Web browser. It has many innovative features (is mostly declarative). I will need a compiler writer soon enough, would like to discuss this project with you.

    • I’m not available at the moment as I already took a full-time job offer, but I’d be happy to give you compiler design advice if you need any. You can message me on twitter @Love2Code or on LinkedIn if you want to exchange contact information.

  5. Mark Schwarzmann permalink

    Hi Maxime. I just wanted to point out that your Gravatar profile is somewhat out of date: “Currently pursuing a Ph.D. degree”? :)

  6. Impressed with your Gradient project and work on artificial life in Javascript. Was trying to understand it at a deeper level but new to genetic algorithms, wondered if you had made a tutorial or blog post unpacking that project more deeply, showing how all the pieces fit together etc. Would be very helpful. Thanks

  7. yingying permalink

    Hi Maxime, I am a new Bee in compiler design but I love it. How can i do something brilliant(higgs & zetaVM) like you in this field? Can you give me some advice or your email? Thank you sooooo much~

    • I would recommend that you start by learning as much as you can about compilers, virtual machines, assembly language, etc. You should know how to program in C/C++. Take compiler classes in university if you can, or watch online lectures from Stanford/MIT, many of them are on YouTube.

  8. Hello Maxime,

    We would love to invite you to be a guest on the XX Files radio show on CKUT 903FM (Montreal). but can not find your email address. The show is about technology from a feminist perspective. Please let us know if this would interest you and if so, how to reach out.

    Thank you!

    best,

    Julia and Amanda

  9. segfault permalink

    I found the site through a random websearch and since I liked the content, I’ll be soon writing a fancy pants wget one liner. Is there any special meaning behind the name?

  10. Cary Swoveland permalink

    I just read, “Ruby 3.2’s YJIT is Production-Ready” and was mightily-impressed, not only with the content but with the attention to detail and the word-smithing. It looked like it had gone though many, many edits.

Trackbacks & Pingbacks

  1. Typed Shapes Paper Submitted | Pointers Gone Wild
  2. Zeta’s JITterpreter | Pointers Gone Wild

Leave a comment