Gauss is an independent research lab working across machine learning, reinforcement learning, and operations research, turning hard theoretical questions into algorithms and systems that learn, plan, and decide under uncertainty.
We work at the meeting point of learning and optimization, developing the theory and the systems that let machines reason about uncertainty, act under constraints, and improve from experience.
Connected lines of inquiry spanning learning, sequential decision-making, and the mathematics that underpins them.
Representation learning, probabilistic models, deep architectures, and the statistical foundations of generalization from data.
Sequential decision-making, policy optimization, exploration, multi-agent learning, and control under uncertainty.
Mathematical optimization, combinatorial and stochastic programming, scheduling, and large-scale decision models.
Bayesian inference, uncertainty quantification, causal reasoning, and robust decisions under noise and partial information.
Where learning meets operations research, neural combinatorial optimization, differentiable solvers, and data-driven heuristics.
Interested in a joint research direction?
Start a conversation →Every line of work follows the same disciplined arc, posed precisely, formalized carefully, validated against evidence.
Identify the open problem, its assumptions, and what a meaningful answer would look like.
Cast the problem mathematically, variables, objectives, and the structure of uncertainty.
Design algorithms and models, grounded in theory and prior work.
Test against benchmarks and ground truth, and quantify what was gained.
Publish findings openly and turn them into systems others can use.
Representative problems and domains where our methods make a difference.
Gauss exists to turn hard questions in learning and optimization into theory, algorithms, and open systems that others can build on.
We work where mathematics meets computation, equally at home in a proof, a benchmark, and a production codebase.
Methods chosen for correctness, not novelty, grounded in statistics and optimization theory.
Ideas that run, tested, reproducible, and built to live beyond the paper.
We share methods, code, and results so the work can be checked, reused, and extended.
Fluent across ML, RL, and OR, one lab connecting ideas that are usually kept apart.
Translating between mathematical depth and real-world problems, without losing either.
Share a short description of what you are working on. We will follow up with a considered, technical perspective and clear next steps.
contact@gauss.pt