AI RESEARCH & SECURITY
We investigate how intelligent systems learn, decide,
and behave — and what it takes to make them reliable.
LOOK CLOSER. ASK A DIFFERENT QUESTION.01 / RESEARCH & PRACTICE
Hidden Delta brings together fundamental AI research, security analysis, and the engineering needed to put ideas to the test.
Learning, optimization, and decision-making. We explore reinforcement learning, symbolic methods, and interpretable models, with particular interest in continuous and dynamic environments.
Reinforcement learning · Symbolic regression · Optimal control
Understanding how systems fail is part of understanding how they work. Our focus spans the security of AI systems and the use of AI in security research, including adversarial evaluation, software analysis, and trust boundaries.
Adversarial testing · Model robustness · Software & API security
Turning open questions into experiments and working systems. We connect mathematical modeling, simulation, and data analysis with practical software and research infrastructure.
Numerical optimization · Simulation · Research software
02 / OUR APPROACH
A useful result starts with a well-posed question. We work from explicit assumptions, test against meaningful baselines, and examine where a method breaks down.
Define the problem, the objective, and the limits of what is known.
Build models, run experiments, and test competing explanations.
Translate the findings into methods, software, and clear technical decisions.
Our research foundations include symbolic regression for reinforcement learning in continuous spaces — connecting interpretable mathematical expressions with sequential decision-making.
Research collaborations, technical challenges,
and ideas worth investigating.