Carina Hong’s path into artificial intelligence began far earlier than her decision to leave Stanford University. Before founding Axiom Math in March, Hong studied mathematics and physics at the Massachusetts Institute of Technology, graduating with a grade of 4.9 out of 5.0. She then completed a Master of Science in Neuroscience at the University of Oxford. Her academic record marked her as a rising researcher, and she later received a Rhodes Scholarship. Yet she chose to step out of formal study and move into startup building, driven by a belief that the next frontier in AI will be shaped by mathematical reasoning.
Building Axiom Math and setting an ambitious technical goal
Axiom Math is Hong’s attempt to build an AI system that can handle advanced mathematics. According to Business Insider, the company recently stated that it had solved two Erdös problems that had challenged researchers for decades. The announcement came shortly after Axiom secured a $64 million seed round and began assembling a small but dense team of researchers. The company currently has seventeen employees, many of whom have joined from Meta’s Fundamental Artificial Intelligence Research (FAIR) lab and its GenAI team. Others have arrived from Google Brain, which merged into DeepMind in 2023.
Why Meta researchers followed her
Hong told Business Insider that one reason she was able to attract senior scientists from large technology firms was the nature of the problem she is trying to solve. In her words, recruits saw the pursuit of mathematical superintelligence as work that could define their careers. This appeal allowed her to approach groups known for deep research output, including Meta’s Fundamental Artificial Intelligence Research laboratory, a unit recognised for long-term inquiry. The laboratory experienced layoffs in October and later saw its chief scientist, Yann LeCun, announce his departure to start his own company. The movement of senior researchers created space for Axiom to hire from teams that had been central to Meta’s artificial intelligence work.
The early team and the talent that followed
Among the first to join Axiom was Shubho Sengupta, whom Hong met by chance in a coffee shop, Business Insider reports. He is now the company’s Chief Technology Officer. Other researchers who have moved from Meta to Axiom include Francois Charton, Aram Markosyan and Hugh Leather. Hong said she was not aware of the specific retention offers Meta may have made while she was recruiting, but she noted that Axiom’s long-term potential was a significant factor for those considering a move.The company’s early days, as described by Hong, began with a folding table and a borrowed couch. Yet the mission attracted researchers from both industry and academia. The Wall Street Journal reported that Hong also hired her former professor, the mathematician Ken Ono. Hong has said that working with senior colleagues is familiar to her from her time in academic research and that age and experience are concepts she treats as flexible. She has aimed to give Axiom a culture that is not hierarchical.
A mission that reaches beyond mathematics
While Axiom’s immediate focus is mathematical reasoning, Hong has pointed to broader applications. She told Business Insider that provably correct reasoning could support areas such as hardware verification, software verification, quantitative finance and cryptography. Her work suggests an ambition that stretches beyond single breakthroughs and toward building systems that can support complex decision-making.
What her journey signals
For now, Axiom is a young company with a small team and a large problem to solve. Yet Hong’s ability to bring in senior researchers from Meta and DeepMind has shown how ambitious ideas can shape talent flows in artificial intelligence. Her path from Massachusetts Institute of Technology to Oxford and then away from Stanford University reflects a view that the most significant work often sits outside established routes. The question now is whether Axiom’s pursuit of mathematical intelligence can match the scale of its early promise.
