About Me


I am currently a Research Scientist in Natural Language Processing at Nuance Communications. My role consists of developping, improving and maintaining deep learning NLP model. Additionally, I have a M.Sc. degree in rehabilitation robotics from Polytechnique Montreal. My experience and technical interests include robotics, controls, computer vision, artificial intelligence and autonomous vehicles (I am also a Space nerd!).

With a strong mathematical and programming background, my main technical skills include Python(opencv, sklearn, numpy, pandas, etc.), deep learning with Keras and Tensorflow, C/C++, Matlab/Simulink, as well as Robot Operating System (ROS) and various other tools.

I have had a chance to participate in multiple projects since I started studying engineering, some of them using the Agile method. I'm a dedicated and conscientious engineer, with strong leadership and teamwork skills and a profound desire to learn and improve.

Outside of engineering, I also enjoy doing multiple things: I play volleyball and do calisthenics multiple times a week, I enjoy the outdoors and going hiking, I play drums in my offtime, and I particularly enjoy movies and books (Ask me about my favorites!).

Completed projects(Click to see details)

"I'm a dedicated and conscientious engineer,
with strong leadership and teamwork skills and
a profound desire to learn."

Work Experience


Research Scientist in NLP 2020-Present

Research Scientist at Nuance Communications. My role is to contribute to the improvement of Nuance's NLP solutions, particularly for chatbots. My main contributions include optimizing parameters to improve performance, as well as helping to maintain and update codebases.

In particular, most of my work has been on improving model performance on very small training sets without altering the model architecture. In short, I am looking at different techniques (simple data augmentation, complex data augmentation using generative models, Adversarial Training, Few-shot Learning, etc.).

Research Development Intern Winter 2020

Intern at Nuance Communications. My role for this internship was to contribute to the development of a proof of concept for Nuance's new computer vision projects. I worked on implementing state-of-the-art skeleton-based human pose recognition methods, as well as developping a method for human head pose estimation. For this project, I had to go through the whole design methodology for a deep learning problem: literature review, data annotation, data augmentation, implementation, transfer learning, debugging, optimization and development with HPC.

Intern - Software developper Summer 2018

Intern at Analogic Canada. My role consisted of developping image analysis algorithms to detect defects in X-Ray detectors using Python, as well as to translate legacy C++ code to Python and vice-versa. Furthermore, I automated the approval process of an X-ray detector following test results to speed up production. I also participated in organizing and managing the project using Scrum, as well as Sprints with weekly team meetings.

Research Intern in robotics September 2015 - May 2018

During my undergraduate studies, I was working part-time at the Research Group in Design, Machine Learning and Optimization for Mechatronic systems (COSIM) at Polytechnique Montreal. I worked on multiple projects, including the control of a robotic arm using facial recognition and the optimization of design parameters for the robust design of a quadrotor (drone). During the summer of 2017, this internship was complete in Melbourne, Australia, in collaboration with the Royal Melbourne Institute of Technology (RMIT).

Research Intern in Biomedical Imaging Summer 2015

After my first year of engineering, I had the opportunity to work at the Laboratory of Optical Diagnoses and Imagery (LODI) with Prof. Caroline Boudoux. My project consisted of designing a variable length reference arm for an Optical Coherence Tomography (OCT) system in order to reduce the noise in images without the use of slow software correction.

Member of PolyProject (Engineering student club) 2014-2018

The main goal of this student club is to complete a different innovative project every year. In my first year, the goal was to build a human-like robot arm and hand. The hand designed was only missing 3 of the human's hand Degrees of Freedom (DoF). In my second year, I had to work on the design of an interactive control system for the robotic hand. The prototype used a sensory glove using fiber optic in order to follow the hand's movements. In my third and fourth years, we worked on designing an intelligent system capable of monitoring the electricity usage in a standard home.

Additionally, I had the opportunity to fill a more administrative role in PolyProject.

  • Treasurer (2016-2017)
  • Director of Communications (2015-2016)

Education


After graduating in E.E., I decided to specialize in robotics with a professor I had been working with for over 3 years, Prof. Sofiane Achiche.

After completing all the classes required for my M.Sc (artificial intelligence, computer vision, controls, etc.), I ended up with a 4.00/4.00 GPA



My project was to develop an autonomous feeding system for people with disabilities using the MICO arm by Kinova. The main tasks consisted of:

  • Implementing a food detection algorithm with Tensorflow & Python to classify and locate food in the 3D workspace
  • Setup-ing a ROS environment for simulation and control. This environment is now the baseline used in the lab.
  • Developping an intuitive and robust way of sending commands for people with limited movement

As an Electrical Engineering undergraduate student at Polytechnique Montreal, Canada, I had the opportunity to learn about the following fields:

  • Controls (stability, full-state feedback controllers and observers, discretization, real-time systems)
  • Signal Processing
  • Artificial intelligence and computer vision(classical methods and deep learning)
  • Advanced calculus and probability

I graduated with honors with a 3.84/4.00 GPA and multiple awards.

Awards & Scholarships


  • Research scholarships
    • Master Student Scholarship by the FRQNT (17 500$, 2019-2020)
    • Alexander Graham Bell Canada Graduate Scholarships by the NSERC (17 500$, 2018-2019)
    • Undergraduate Student Research Awards by the NSERC (4500$, 16 weeks in 2017)
    • Undergraduate Student Research Awards by the NSERC (4500$, 16 weeks in 2016)
    • Undergraduate Student Research Awards by the NSERC (4500$, 16 weeks in 2015)
    • Undergraduate Initiation to Research Scholarship by Polytechnique Montreal (1500$, 80 hours in 2015-2016)
    • Undergraduate Initiation to Research Scholarship by Polytechnique Montreal (1500$, 80 hours in 2016-2017)
    • Undergraduate Initiation to Research Scholarship by Polytechnique Montreal (1500$, 80 hours in 2017-2018)
  • Scholarships & awards
    • J.A. DeSève award by Polytechnique Montreal (4000$, 2019)
      • Academic achievements
      • Extracurricular Involvement
    • CMC Electronics award by Polytechnique Montreal (2500$, Won twice: 2017 and 2018 )
      • Academic achievements
      • Extracurricular Involvement
    • Vedel award by Polytechnique Montreal (2000$, 2016)
      • Academic achievements
      • Extracurricular Involvement
    • Hatch Ltd. award by Polytechnique Montreal (2000$, 2015)
      • Academic achievements
      • Extracurricular Involvement
    • Entry Scholarship by Polytechnique Montreal (2500$, 2014)
      • Academic achievements in pre-university education
      • Extracurricular Involvement
  • Others
    • 2nd place at the Polytechnique Engineering Competition (2016)
      • 2nd best performing team at the event (8 hours competition)
      • Design and build a crane with an electromagnet to fetch objects on a workspace, using limited materiels
    • 2nd place at the Polytechnique Engineering Competition (2015)
      • 2nd best performing team at the event (8 hours competition)
      • Design and build a catapult, without electronics, to throw an object as far as possible

Skills


Some of my most significant skills to help in engineering tasks.

Tensorflow/Keras

90%

Python(numpy, pandas, sklearn, etc.)

90%

C/C++

7%

OpenCV

80%

Git (Version control)

90%

Robot Operating System (ROS)

80%

Java

70%

Matlab/Simulink

90%

LaTeX

75%

Controls & dynamics

90%

Cooking Mexican cuisine

110%

MS Office

95%

Contact


If you would like to get in touch for any question, inquiry, comment, or to connect and discuss, feel free to email me at gabriel.descoteaux@polymtl.ca. You can also reach me on LinkedIn.