Machine learning
From framing the problem to training, evaluation, and a model that holds up in production — not just in a notebook. We handle the data, the modelling, and the plumbing that keeps it healthy.
We build machine-learning systems, trading algorithms, data pipelines, and search engines. Every engagement runs through the same four stages — from first spec to power on.
We take on work we can stand behind — scoped honestly, built with rigor, and shipped to production. No hype, no black boxes.
Four disciplines, one standard — production-grade or not at all.
From framing the problem to training, evaluation, and a model that holds up in production — not just in a notebook. We handle the data, the modelling, and the plumbing that keeps it healthy.
Research, backtesting infrastructure, signals, and low-latency execution — strategies built to survive real markets, not just the in-sample period.
Scrubbing, wrangling, and pipelines that turn messy, multi-source data into clean, validated, queryable databases with documented rules.
Search engines, retrieval systems, and AI-powered products — fast, relevant, and built to scale as data and traffic grow.
From first conversation to production, with full visibility the whole way.
Tell us the problem and the goal.
A costed delivery plan within a week.
Iterative, tested, and reviewed as we go.
Shipped to production and monitored.
We build machine-learning systems, trading algorithms, data pipelines, and search engines — and take them all the way to production. We are a software and AI engineering studio, not a solar or hardware company.
Typical work is data cleaning and structuring, a scoped AI/ML pilot on real data with a measurable goal, putting a model into production against your existing systems, and ongoing operation and monitoring.
In fixed-price stages, so there is no open-ended consulting risk: a scoped pilot with written acceptance criteria, then a production step, then optional operation and monitoring. You always know the price up front.
Pilots are measured in weeks, not months. We agree the acceptance criteria before we start, so 'done' is defined in writing from day one.
In Sweden — Göteborg and Stockholm — working with clients across the Nordics, on-site or remote.
EU/EES data handling, a data processing agreement from day one, and no training on client data. AI features are built to meet transparency requirements under the EU AI Act.
Sign in with your work Google account to open the shared project board. Access is limited to approved team members.