Victoria Krakovna

Victoria (Viktoriya) Krakovna is a research scientist in AI safety at Google DeepMind, working to ensure advanced AI systems do what we want them to do and do not knowingly act against our interests. Her research covers deceptive alignment, dangerous capability evaluations, specification gaming, goal misgeneralization, and avoiding side effects. She co-founded the Future of Life Institute, a nonprofit working to reduce technological risks to humanity, and sits on its board of directors. Her Harvard PhD in statistics and machine learning focused on building interpretable models. She competed for Canada at the 2006 International Mathematical Olympiad, winning a silver medal, and won the 2008 Elizabeth Lowell Putnam Prize.

Notes6

• Quant employment verification: CONFIRMED. Her own CV lists 'Quantitative Analyst Intern, D.E. Shaw & Co (2012)', developing and testing risk modeling algorithms using statistical and numerical optimization methods in Python. D. E. Shaw & Co. is a quantitative/systematic investment firm (hedge fund), so the employer operates in quantitative finance; the role is a paid internship, not a student program, competition, or insight day. Identity is supported by the same document and profile (Harvard statistics PhD, IMO silver medalist representing Canada, DeepMind research scientist, FLI co-founder). Dates: 2012 (LinkedIn index shows May-Aug 2012).2012
• Identity resolved: LinkedIn (linkedin.com/in/vkrakovna) is Victoria/Viktoriya Krakovna, Senior Research Scientist at Google DeepMind and co-founder of the Future of Life Institute; LinkedIn displays 'Victoria Krakovna', while Olympiad and press usage is 'Viktoriya Krakovna'.2026
• Career timeline per her own CV: Google DeepMind Research Scientist 2016-present (Senior Research Scientist from 2020); Software Engineering Intern at Google 2015; Decision Support Engineering Intern at Google 2013; Teaching Fellow in Statistics at Harvard 2012-2013; Quantitative Analyst Intern at D. E. Shaw & Co. 2012; Summer Research Analyst in Computer Science at University of Toronto 2009; Teaching Assistant in Mathematics at University of Toronto 2007-2011.2024
• Co-founded the Future of Life Institute (2014) and is a member of its Board of Directors; FLI is a nonprofit working to reduce technological risks to humanity.2024
• Education: Harvard University PhD in Statistics (2016), thesis on building interpretable models; University of Toronto MS in Statistics (2011) and Honors BS with High Distinction (GPA 3.76/4.00) in Statistics/Mathematics (2010).2016
• Research focus: AI alignment at DeepMind spanning deceptive alignment, dangerous capability evaluations, specification gaming, goal misgeneralization, and avoiding side effects; PhD work was in statistics and machine learning on interpretable models.2024
• Competition record before her research career: silver medal for Canada at the 2006 International Mathematical Olympiad (rank 76, 22 points), Elizabeth Lowell Putnam Prize 2008 (highest-ranking woman in the Putnam), University of Toronto Putnam team 2006-2009, and ACM regional programming team 2007-2008; she later served as deputy leader of the Canadian IMO team in 2010.2008
• Public voice/profile hubs she links herself: Twitter/X @vkrakovna, Google Scholar, GitHub vkrakovna, LinkedIn, and the Alignment Forum user 'vika' - a self-consolidated identity set that confirms these handles belong to one person.2024
• Service record: co-organizer/program chair of the Beneficial AGI (BAGI) conference 2019 and organizer of the NeurIPS ML Safety workshop 2022 and ICLR Safe ML workshop 2019; reviewed for ICML, NeurIPS, ICLR and JMLR; sat on the CIFAR International Scientific Advisory Committee for the Pan-Canadian AI Strategy (2018); was named a top 30% NeurIPS reviewer in 2018.2022
• Podcast appearance: AXRP episode 7 (May 2021) on side effects, reflecting her published work on avoiding side effects in RL agents.May 14, 2021
• Harvard dissertation 'Imputation of Possibilistic Data for Structural Learning of Directed Acyclic Graphs' reflects her statistics PhD work on graphical/structural models, a line predating her AI-safety focus.2014