CV

Education

• B.S. in Software Engineering, École Polytechnique de Montréal, 2018
• Ph.D in Deep Learning, MILA / École Polytechnique de Montréal, 2022 (expected)

Research Interest

• Natural Language Processing
  • Exploring emerging properties from very large pre-trained models such as BERT.
  • Zero-shot question answering as well as information embedding within a model’s parameters.
• Meta-Learning
  • Application relating to Meta-Learning and 1D data such as time-series and text.
• Recurrence and Transformers
  • Understanding how non-recurrent approaches to 1D data, such as Transformers, compare with recurrent approaches.
  • Studying the importance of the dimensionality of sequences.

Work experience

• Summer 2018 - ongoing: Quantitative Research Scientist
  • Shell Street Labs
  • Conceptualize, develop and deliver trading strategies used to invest hundreds of millions of dollars
• Summer 2018: AI Research Intern
  • Ubisoft, LaForge
  • Implement and iterate with state of the art deep learning text to speech models
• Summer 2017: Data science Intern
  • Société Générale
  • Developed machine learning based solution to predictive maintenance on a wide array of technologies.

Kaggle

• Kaggle Master
  • Best global rank : 442th / 133469 Data scientist
  • Gendered Pronoun Resolution: Gold Medal
  • Jigsaw Unintended Bias in Toxicity Classification : Silver Medal
  • Sberbank Russian Housing Market: Silver Medal

Publications

FIGR: Few-Shot Image Generation with Reptile

Open Source Project

• PPO-Keras
  • The first implementation of Proximal Policy Optimization in the Keras framework.
• FIGR
  • Few-shot image generation by meta-training a Generative Adversarial Network with the Reptile algorithm.
• FIGR-8
  • Dataset containing 1.5 million images in 17000 classes. Intended for few-shot classification or few-shot image generation tasks.

Perfectly trilingual: French, English and Python