AI consulting
I help engineering and product teams apply AI across the software development lifecycle, and keep it defensible when someone asks hard questions about it.
Most AI work fails in the same three places. The data does not hold up, nobody scoped the operational load, or the system cannot be explained to the people who have to sign off on it.
What I work on
AI in the product
Forecasting, computer vision and LLM features that survive real users and real volume. The model is rarely the hard part. The data behind it, the evaluation that proves it still works next month, and the operational load nobody costed are.
AI in the delivery
Specification, code, review, testing, documentation, operations. Done well this compounds. Done badly it generates more code than your team can review, which is slower, not faster.
AI you can defend
EU AI Act readiness, GDPR, bias and fairness review. The first thing a Norwegian board asks about now, and the part most consultants skip. See algorithmic bias in Norwegian hiring and a working summary of the EU AI Act.
How engagements work
Most start with a short paid review. I look at what you have, talk to the people building it, and write up an assessment with a prioritised plan you can act on, with me or without me.
Longer advisory arrangements follow when they earn their place. I take a limited number, so fit matters more than availability.
Background
Currently at Izy, maintaining the production backend for a Norwegian workplace platform and building the computer vision and demand forecasting that run on top of it. Live in commercial buildings, not in a demo.
Eight years across backend architecture, machine learning and technical product. Distributed services, data pipelines and ETL, most of it before the current AI cycle. Two peer-reviewed papers. 160 hours of formal training in AI ethics. Registered Scrum Master.
Questions
Do you work with companies outside Norway? Yes. Based in Oslo (CET), working remotely with clients internationally, in English and Norwegian.
Do you build, or only advise? Both. The recommendations come from someone who has had to operate these systems, not just specify them.
What does the EU AI Act mean for a Norwegian company? Norway implements it through the EEA agreement, so it applies to Norwegian companies placing AI systems on the European market. The work is classifying your systems by risk tier, then producing the documentation and evaluation that tier requires.
Tell me what you are building. Book a conversation or email post@saurav.no.