Machine Learning Fellow - Human Frontier Collective (Canada)

Scale AI · Human Frontier Collective

About the Collective The Human Frontier Collective (HFC)’s mission is to bring the world's best minds to shape the future of AI – because the future of AI depends not only on powerful models, but on the people who teach them to think. We bring together researchers, academics, and domain leaders across 70+ fields – over 90% of members hold doctorates – to shape how frontier AI systems are built, evaluated, and governed. HFC members have since co-authored published work including SciPredict, PropensityBench, and Professional Reasoning Benchmark. Why join the HFC HFC centers on three core pillars: the network itself, participation in selected AI/ML projects, and opportunities to publish with our research team. Join a unique and exclusive expert network: You’ll become a part of an interdisciplinary community, consisting of over 90% doctorates and field leaders representing 70+ fields from the leading institutions. We’re a collective of top innovators and thought leaders committed to advancing frontier AI to power the world’s most important decisions. Participate in AI/ML projects: Beyond the collective network, you’ll be regularly invited to work on high-impact projects with Scale AI and its affiliated lab and platform: building AI safety and policy guardrails, helping models understand real-world deep learning workflows by designing, reviewing, and optimizing PyTorch models, evaluating complex ML code and AI-generated implementations for efficiency and correctness, and more. Co-author selected research publications: Collaborate with Scale’s research team to co-author technical reports and research papers—boosting your academic visibility and professional recognition. Who should apply Background: PhD, or postdoctoral research experience, in Machine Learning, Computer Science, or a related field; or senior industry experience in Machine Learning or a related field. Skills: Strong proficiency in Python and modern ML frameworks (PyTorch, TensorFlow). Experience with cloud infrastructure (AWS) and MLOps tools (Docker) and LLM frameworks (LangChain) is a plus. Professional Mindset: Detail-oriented, innovative thinker with a passion for applied AI research and a commitment to collaboration. How it works Duration: This is a fully remote opportunity with no fixed end date; engagements continue as long as there's mutual interest. This is a 1099 independent contractor engagement. Work authorization: We do not sponsor visas for Fellows. To participate in the HFC Fellows program, you need to have or independently obtain full-time work authorization in the country they reside in. Community Network: Once you’ve received an invitation to join the HFC, you’ll also gain access to our newsletters, discussion channels, virtual sessions, IRL events, and much more. Projects Logistics: Project selection: You’ll regularly get matched to carefully selected projects from our partners. Depending on the partner, projects range from evaluating AI models to designing experiments, building RL environments, co-authoring research papers, and more. Flexible schedule: There's no minimum commitment – you decide whether to take on a project. Most fellows spend 10–25 hours per week, on a schedule they set themselves. Competitive pay: Project pay rates vary across platforms and depend on a number of factors, including but not limited to: projects, scope, skillset, and location. You’ll receive the pay information upon receiving project matching notifications. Application process Apply: We review applications on a rolling basis. Interview: Candidates will get to discuss their research experience, professional background, and alignment with our mission to advance human-centered AI. Note: If we invited you directly, your invitation will say whether this step applies. Join the Collective: Successful candidates will receive an invitation to join the Human Frontier Collective Fellowship. ------ PLEASE NOTE: Our policy requires a 90-day waiting period before reconsidering can