42 is an independent AI lab building foundation models that perceive, research, decide, and act in real-world environments. Financial markets serve a dual purpose: the harshest proving ground for the company's research and the engine that funds it.
We are working on:
Multimodal world models (time series + text + structured data)
Decision-making under uncertainty (not just prediction)
Models that optimize for realized outcomes (PnL, risk, execution)
Online / continual learning in non-stationary environments
Autonomous agents that plan, act, and learn from interaction with the environment
Research agents that generate hypotheses, design experiments, and iterate without human supervision
Most labs optimize benchmarks and publish static results in controlled settings. We build world models embedded in agents that act in real environments, learn from continuous feedback, and produce real outcomes — not just papers.
Research at 42:
Led by Irina Rish, Professor at the Université de Montréal (UdeM) and a core faculty member of MILA - Quebec AI Institute
Small, high-density team of researchers and engineers
No dependency on users, growth, or sales — research is driven solely by improving the system’s capabilities and real-world performance.
You might fit if you:
care about building systems, not just papers
think in terms of models of the world
are frustrated by static benchmarks
want your research to interact with reality
42 is an independent AI lab building foundation models that perceive, research, decide, and act in real-world environments. Financial markets serve a dual purpose: the harshest proving ground for the company's research and the engine that funds it.
We are working on:
Multimodal world models (time series + text + structured data)
Decision-making under uncertainty (not just prediction)
Models that optimize for realized outcomes (PnL, risk, execution)
Online / continual learning in non-stationary environments
Autonomous agents that plan, act, and learn from interaction with the environment
Research agents that generate hypotheses, design experiments, and iterate without human supervision
Most labs optimize benchmarks and publish static results in controlled settings. We build world models embedded in agents that act in real environments, learn from continuous feedback, and produce real outcomes — not just papers.
Research at 42:
Led by Irina Rish, Professor at the Université de Montréal (UdeM) and a core faculty member of MILA - Quebec AI Institute
Small, high-density team of researchers and engineers
No dependency on users, growth, or sales — research is driven solely by improving the system’s capabilities and real-world performance.
You might fit if you:
care about building systems, not just papers
think in terms of models of the world
are frustrated by static benchmarks
want your research to interact with reality
42 is an independent AI lab building foundation models that perceive, research, decide, and act in real-world environments. Financial markets serve a dual purpose: the harshest proving ground for the company's research and the engine that funds it.
We are working on:
Multimodal world models (time series + text + structured data)
Decision-making under uncertainty (not just prediction)
Models that optimize for realized outcomes (PnL, risk, execution)
Online / continual learning in non-stationary environments
Autonomous agents that plan, act, and learn from interaction with the environment
Research agents that generate hypotheses, design experiments, and iterate without human supervision
Most labs optimize benchmarks and publish static results in controlled settings. We build world models embedded in agents that act in real environments, learn from continuous feedback, and produce real outcomes — not just papers.
Research at 42:
Led by Irina Rish, Professor at the Université de Montréal (UdeM) and a core faculty member of MILA - Quebec AI Institute
Small, high-density team of researchers and engineers
No dependency on users, growth, or sales — research is driven solely by improving the system’s capabilities and real-world performance.
You might fit if you:
care about building systems, not just papers
think in terms of models of the world
are frustrated by static benchmarks
want your research to interact with reality
