This program is for candidates who are passionate about Artificial intelligence and offer diverse perspectives.
How to apply
Click on the links below to find out more about our Intern positions and apply.
Deep Learning and Language
The mission of Microsoft Research Montréal’s Deep Learning and Language Team is to build machines that learn from and understand the world, with a particular focus on understanding through language. We are a leader in using deep learning to solve complex language-understanding problems and in training machines to model reasoning and decision-making capabilities. Based in Montréal, a global hub for Artificial Intelligence (AI), our team brings together experts in various machine-learning domains and collaborates deeply with the surrounding academic community.
Natural Language understanding with Deep Learning
Specifically, for this Internship, we are interested in collaboration on deepening our understanding of representation learning on the lexical level, including:
- How to improve model generalization on rare words, particularly by leveraging meta-learning
- How to better model compositionality from the lexical level up
- How to accommodate for non-compositional exceptions in data to improve model robustness
- How does representation evolve from being context-independent (on the leixcal level) to being contextual
- How to leverage this evolution to improve representation learning and/or model efficiency
Offline Reinforcement Learning
The goal of this project is to develop fundamental methodologies with formal grounds to enable reinforcement learning from offline data with no prior knowledge about how the data is collected. This is a particularly important missing step to facilitate using RL in critical domains of special importance, such as healthcare. How the optimal value function can be best learned and where the learned values can be trusted are what we would be aiming for!
The Microsoft Research Montreal Fairness, Accountability, Transparency, and Ethics (FATE) team seeks to align the uses of computational systems to broader societal goals in the service of social good, justice, and fairness. Our aim is to understand the limits of computational systems and their impacts on individuals, groups, and societies. To these ends, we develop quantitative and qualitative techniques to guide the design and application of these systems. We achieve these goals through collaborative, interdisciplinary research.
Analyzing Contrastive Representation Learning
The mission of Microsoft Research Montréal’s Deep Learning and Language team is to build machines that learn from and understand the world, with a particular focus on understanding through language. We are a leader in using deep learning to solve complex language-understanding problems and in training machines to model reasoning and decision-making capabilities. Based in Montréal, a global hub for AI, our team brings together experts in various machine-learning domains and collaborates deeply with the surrounding academic community. The Deep Learning and Language team explores representation learning, natural language understanding, generation, and the use of language as a tool for acting in the world.
Representation Learning for Reinforcement Learning
This internship will be working in collaboration with the Reinforcement Learning and the Deep Learning and Language teams at Microsoft Research Montréal. It will focus on representation learning in the context of reinforcement learning.
Reinforcement Learning Actor-Critic Algorithms
This internship project will focus on Actor-critic algorithms in RL, specifically investigating the theoretical validity of various discounting schemes and their applicability to new algorithms.
Causal Priors in Reinforcement Learning
This internship project will focus on trying to improve generalization performance of RL through causal priors. Specifically, the intend is to study how to encode causal priors into representation learning (combining contrastive learning, bisimulation and other useful learning constructs) in a multi-task setting.
Parametric uncertainty for Offline Reinforcement Learning
This Internship will be focusing on Safe Policy Improvement (SPI) in Batch Reinforcement Learning (Batch RL). Our aim is to solve the limitations of SPIBB algorithms by using parametric uncertainty estimates to evaluate how much a given state has been visited within a given dataset and investigate new techniques to solve the problem at hand.
Leveraging knowledge sources to solve tasks efficiently
This Internship will be focusing on using knowledge sources to enable the agent to solve tasks more efficiently. We expect intelligent machines to develop intricate skills as querying external information relevant to the task at hand and are exploring the “Big Questions” relating to this area of research such as:
- Can an agent learn to ask questions to reduce the amount of exploration needed?
- Can an agent learn to ask questions to avoid dangerous states?
- Can an agent learn to query other agents to better collaborate?
Being an Intern at MSR Montreal
During the 12-week internship, you will be supporting Microsoft with its mission of empowering every person to achieve more.
We don’t just value differences, we seek them out. We invite them in. We are a team of individuals at a truly global company who work to ensure our interns feel valued, respected, included, and encouraged.
There are great opportunities for you to build upon your skills and experience with us as you will have the opportunity to collaborate and learn from our world leading research team who are dedicated and invested in helping you.
You will be working on strategically important research areas that will shape and drive innovation within the AI space and be given intellectual freedom whilst working in a supportive environment.
Build valuable relationships
“Getting to interact with Yoshua Bengio about the project was the highlight of the internship”.
We offer multiple opportunities to connect, network, and build lasting relationships throughout the intern program through a variety of collaboration and networking events, workshops, and research talks.
You will be an important part of a team, and your mentor is dedicated to ensuring you to get the best out of your internship and will take the time to help you and discuss your ideas.
“I attended meditation sessions, a cocktail class, a tri-lab speed networking event, all were great fun.”
“Receiving little packages in the mail with supplies for the events brightened my days during the lockdown in a way that I didn’t expect and truly appreciated.”
We hold a number of events and activities during the summer to enable you to get to know others and try something new while enjoying what MSR Montreal has to offer.
Competitive pay and the opportunity to build your career with Microsoft
We offer a competitive salary and an accommodation allowance. Many of our interns have gone on to get full-time positions here or at other Microsoft locations.
Our internships are exclusively offered to PhD students.