Research Associate

McGill University • Montreal, QC

8d • Onsite

About The Position

We are seeking a highly skilled and motivated Research Associate to lead the technical development and management of neuro-foundation models. Working closely with the research team at Mila and the McGill community, the successful candidate will play a critical role in developing our specialized public codebase (NeuroGalaxy) and supporting the broader open-source community.

Requirements

• A PhD in neuroscience, computer science, or a related field.
• More than 2 years of proven experience with creating, releasing, and maintaining large-scale open-source code bases.
• Familiarity with the most common types of neural data file formats (e.g., NWB, NIDS, etc.).
• Extensive experience training AI models on high-performance GPU clusters.
• Knowledge of French and English: McGill University is an English-language university where day to day duties may require English communication both verbally and in writing. The level of English required for this position has been assessed at a level 4 on a scale of 0-4.
• Authorized to work in Canada and willing to work in the province of Quebec at the campus where the position is based / located.

Responsibilities

• Codebase Management: Overseeing the continued development and maintenance of a large, specialized public code base for creating neuro-foundation models (NeuroGalaxy).
• Quality Assurance: Debugging the code base, managing the GitHub repository, and approving new pull requests from the community of users.
• Technical Documentation: Creating robust, accessible documentation for the code base, as well as public-facing websites and tutorials for new users.
• Resource & Grant Support: Providing estimates of compute costs associated with different components of the code base for different model and dataset sizes. Writing specialized technical analyses of existing compute usage on the DRAC and Mila clusters for use in grant applications.
• Model Analysis & Integration: Analyzing new models generated by the community and incorporating their designs into the codebase.
• Benchmarking & Training: Training models created by the Mila community, benchmarking them against existing models, and maintaining a list of critical benchmarks alongside a leaderboard.
• Collaboration: Liaising with other neuro-foundation model development groups in academia and industry to foster partnerships and track advancements.

What This Job Offers

• Job Type: Full-time
• Career Level: Senior
• Education Level: Ph.D. or professional degree