A part-time, fully remote research fellowship that pairs aspiring AI safety and policy researchers with working professionals on projects spanning safety, policy, security and interpretability. Undergraduate, graduate and PhD students as well as professionals at any experience level are eligible, and previous research experience is not required. Participants commit between 5 and 40 hours a week over a roughly three-month research period ending in a Demo Day. SPAR covers project expenses such as compute and API credits but does not pay mentees a stipend.