EQE (Engineered Quotient Evolution) is an AI research lab developing next generation AI systems that autonomously learns, reasons, and generalizes solutions across novel, complex and unpredictable environments.

Despite advances in recent years, current AI remains limited by brittle reasoning, reliance on massive datasets, and energy-intensive computation, creating risks that threaten societal stability, ethical standards, and the environment.

We are building superintelligence based on a scientifically feasible path to counter these risks, serve humanity, uncover new scientific principles, solve previously intractable challenges, and drive transformative advancements that propel civilization toward its next great leap.

We are operators and world-renowned academic leaders who have dedicated our careers to these challenges and have come together to this end. We designed and commercialized ultra-efficient AI hardware, pioneered efficient RNN scaling techniques, and played a pivotal role in formalizing problem definitions for continual learning.

Autonomous Learning

Current AI makes closed-world assumptions and reasons within the limits of training data (interpolation), which makes it brittle in real-world environments. We humans, by contrast, are masters of extrapolation: we form abstract mental models, understand cause and effect, adapt and reason creatively in new environments and across domains.

We close this gap by transitioning from interpolation to extrapolation (Reasoning Beyond the Known) through an architecture that natively supports self-directed, fully autonomous continual learning, allowing AI to build an evolving model of the world capable of uncovering new knowledge and safely integrating it to enable advanced, self-improving reasoning.

Human-Like Reasoning

Humans reason effectively through recurrent processing, which supports reflection (the ability to think about thinking) and continuous refinement of understanding, enabling pattern recognition, advanced reasoning, and context-aware decision-making.

Similarly, our architecture relies on recurrent processing combined with episodic memory, time-based planning, and multimodality, scaled to unprecedented levels to enable advanced reasoning and autonomous continual learning.

Computational Efficiency

We adopt a software/hardware co-design approach, leveraging hardware-aware algorithms to achieve unprecedented compute efficiency for training and inference, enabling us to develop larger, more capable reasoning models that far surpass current industry standards in reasoning capabilities.

This efficiency dramatically lowers compute costs, enabling sustainable business models and products that solve tangible world problems rather than contributing to societal harm. It makes advanced intelligence accessible to all, so AI can be applied where it matters most.

Join Us

Team: Prof. Christian Mayr · Dr. Anand Subramoney · Prof. Bing Liu · Ayman Mackouly · Jamie Taylor.

We are expanding our team. If the above resonates with you, reach out at apply@eqe.ai.

For research leadership roles, candidates must have advanced academic credentials beyond the PhD level; industry experience is a big plus.

For other inquiries, contact comms@eqe.ai.