Senior Machine Learning Research Scientist

Role Overview

We are looking for a Senior Machine Learning Research Scientist to contribute to pioneering research within our AI safety initiative. This position focuses on creating and refining advanced ML models addressing critical AI safety challenges.

Key Responsibilities

• Conceptualize, design, and develop innovative ML models aimed at solving intricate AI safety issues.
• Work closely with mathematicians and domain experts to translate theoretical progress into viable ML algorithms and tools.
• Modify and optimize existing cutting-edge models to suit specialized application contexts.
• Develop rigorous experimental setups and evaluation systems to test hypotheses and guarantee replicable and robust outcomes, including for complex frontier models.
• Analyze experimental results to inform and direct forthcoming research priorities.
• Collaborate with ML engineers to benchmark and enhance model efficiency and performance, ensuring effective resource utilization during extensive computational experiments.
• Effectively communicate research insights and synchronize with cross-disciplinary teams of researchers and engineers.

Required Qualifications

• Advanced degree in Computer Science, Mathematics, or closely related discipline; PhD preferred, though exceptional experience may be considered in lieu of a doctorate.
• At least 5 years of experience undertaking deep learning research, especially involving advanced frontier models.
• Demonstrated proficiency in training, adapting, and fine-tuning frontier ML models for varied complex use-cases such as transfer learning, domain adaptation, and meta-learning.
• Expert knowledge of ML frameworks including PyTorch, TensorFlow, or JAX, particularly in distributed computing contexts for model development and evaluation.
• Strong programming skills in Python, familiarity with version control systems, collaborative development environments, and tools for managing and tracking machine learning experiments.
• Excellent communication abilities, both verbal and written, facilitating collaboration with experts across scientific fields.
• Capacity to work independently and as part of a dynamic team in a collaborative setting.

Preferred Experience

• Expertise in examining the behavior of frontier models under safety-sensitive scenarios.
• Background in designing model alignment techniques or benchmarking frameworks.
• Experience with natural language processing and probabilistic graphical models.
• Proven record of participation in high-impact deep learning research projects.

What We Offer

• An opportunity to contribute to a high-impact mission focused on safe AI development.
• Comprehensive healthcare benefits.
• At least 20 paid vacation days per year from the commencement of employment.
• A mandatory employer contribution to retirement savings of at least 4%.
• Flexible benefits package supporting employee well-being.
• A collaborative environment with a team of passionate experts.
• Workplace located in the vibrant Little Italy neighborhood, close to public transit in Montreal's Mile-Ex district.

About LawZero

LawZero is a non-profit organization pioneering the development of AI systems designed with safety at their core. Guided scientifically by Professor Yoshua Bengio, one of the most cited AI researchers worldwide, LawZero's research tackles safe, non-agentic AI that primarily understands rather than acts upon the world. These AI technologies aim to enhance scientific advancements, oversee agentic AI systems, and deepen understanding of AI risks and mitigation strategies. LawZero's mission is to foster AI as a global public good, engineered responsibly to promote human flourishing.

Commitment to Diversity and Data Privacy

LawZero promotes an inclusive, fair, and respectful workplace valuing diverse perspectives. We encourage applications from highly qualified individuals committed to advancing this mission in a collaborative environment. Applicant data will be processed in accordance with our Privacy Policy, allowing individuals to exercise their privacy rights under applicable legislation.