Viktor SuchýCTO
Viktor started analyzing oil futures at fifteen and has been building systems around data ever since. He owns Euler's architecture: the data pipelines, the models, and the infrastructure they run on.
suchy@euler.czWe build pricing and prediction engines. And trade our own capital by the same rules.
From raw data to production. We use AI where it earns its place, the simplest model everywhere else, and we measure all of it.
n = 1 250
ρ̂₁₂ = 0.00
σ̂ = 0.00
‖z‖ > 2σ : 0.0 %
Trusted by
and more
What we do
We specialize in three areas. In each, we go from the question the business is asking to a system running in production.
- 01
Pricing and valuation
Pricing engines, valuation models, and monitoring of the gap between market and contract prices. Calculations that can be explained and defended in an audit.
Škoda Auto · Hengst · Česká spořitelna / aConsult
- 02
Prediction and data
We pull scattered data into one place and build models that show what drives the result. Often a regression is enough.
FutureSales · ČMKBK
- 03
AI that can be measured
Assistants over company data, document reading, agents that coordinate teams. Always with evals. A language model only where something simpler won't do.
eDO · Škoda Auto · InnovationPath
About
We are a small team of statisticians, econometricians and engineers in Prague. We met at university and still keep close ties there, which is how we stay current with the methods before they reach the mainstream. A model doesn't end for us in a presentation - we take it into production, watch how it behaves over time, and stand behind what it computes. Meet our founders below.
Lukáš TahotnýCEO
Lukáš starts from the business: what a company needs to decide, which number would prove the system worked, and what that is worth. He built and closed a product called BeeSync. He owns strategy and day-to-day operations.
tahotny@euler.czOur values
The simplest system that works. And the numbers to prove it.
Decision first, model second.
We start from the question the business needs answered and the number that will show whether the answer worked.
Complexity has to be earned.
Regression before a neural network, a rule before a language model. A simple system is cheaper, easier to explain, and lasts longer.
Every system has its number.
Forecast error, eval pass rate, hours saved. Without measurement we don't know whether the system works, and neither does the client.
Done means in production.
A model in a slide deck doesn't count. We deliver the pipelines, the application and the monitoring, and we stay when the data changes.
When the data can't answer the question, we say so before anyone makes a decision based on it.