cv
General Information
| Full Name | Lucas Pagé-Caccia |
| Languages | English, French |
Education
-
2018-2023 PhD, Computer Science (fast-track from Masters)
McGill University & Quebec AI Institute (Mila) - Thesis "Preventing Forgetting and Promoting Transfer in Continual Learning"
- 4.0/4.0 GPA
-
2014-2017 B.A. Mathematics & Computer Science
McGill University - 3.96/4.0 GPA (Dean's Honor List)
Experience
-
2022 - 2023 Post-Doctoral Researcher
Microsoft Research - Building scalable and modular systems for large-scale multitask adaptation.
-
2022 - 2023 Machine Learning Research Intern
Microsoft Research - Working on modular approaches for efficient forward transfer in Natural Language Processing.
- Designed better routing mechanisms for soft MoE architectures.
-
2020 - 2022 Visiting Researcher
Meta AI Research (FAIR) - Developed new continual learning algorithms designed for realistic settings.
- Developed an algorithm to perform online compression from non-iid data.
Open Source Projects
-
2022-now Multi-Task Transfer Learning (MTTL)
- Library for parameter-efficient multitask transfer learning
-
2018 Pixel-CNN
- Pytorch Implementation of OpenAI's PixelCNN++
-
2017 Deep Lidar Generation
- Code for "Deep Generative Models for LiDAR Data"
Teaching
-
African Institute for Mathematical Sciences (AIMS)
- Teaching Assistant, for a two week long Reinforcement Learning class held in Kigali, Rwanda. We prepared several tutorials covering the basics of RL (which can be found here).
-
McGill University
- Teaching Assistant, COMP 551 - Applied Machine Learning Class. I gave a few tutorials on the basics of automatic differentiation in Pytorch, and held office hours for students.
Honors and Awards
-
2020-2021 - Borealis AI Fellowship - Selected alongside 10 fellows for research and academic achievements
-
2016-2019 - CIBPA Foundation Bursary (Merit) - Awarded to promising Canadian students of Italian descent
-
2015-2016-2017 - Dean's Honour List - Awarded to the top 10% performing students