AI Engineering • Based in Norway • Available Worldwide

Your AI prototype worked. Now make it production-ready.

We help teams close the gap between a promising demo and a system their users, clients, and regulators can actually rely on.

Start your project → Sound familiar?

Research & collaboration background

NTNU St. Olavs Hospital

Sound familiar?

Most AI projects fail between demo and deployment.

01

The demo impressed everyone. Production is a different story.

Hallucinations, latency spikes, and reliability gaps keep delaying launch. What worked in a notebook breaks under real user load and messy data.

02

AI is a strategic priority, but no one owns the build.

Your engineers are already stretched thin. You need someone who can take an AI initiative from architecture to deployment without pulling your team off their roadmap.

03

A black box won't pass review.

In healthcare, finance, or any regulated space, stakeholders and regulators need to understand why your model made a decision — not just that it did.

04

Every vendor promises AI magic. You need an honest assessment.

Before committing budget and engineering time, you need a clear-eyed evaluation of what AI can actually solve for your specific problem — and what it can't.

How we solve it

From strategy to shipped.

01

AI & ML Consulting

You know AI could help, but you're not sure where to start — or whether the approach you've chosen will hold up. We cut through the noise: assess feasibility, design the right architecture, and give you a deployment plan that accounts for real-world constraints. No vague strategy decks. Concrete technical direction.

FeasibilityArchitectureMLOpsEvaluation
02

RAG & LLM Systems

Your retrieval pipeline hallucinates. Your agent loops. Your context window costs are spiralling. We rebuild it: production-grade RAG and multi-agent systems designed to be reliable, fast, and measurable — not just impressive in a demo.

RAGLangChainLangGraphMulti-AgentFastAPI
03

Explainable & Healthcare AI

Your model performs well on paper, but clinicians don't trust it and regulators can't audit it. We make complex models interpretable — using proven techniques like Grad-CAM for visual explanations — so your AI meets both clinical standards and EU AI Act requirements.

Grad-CAMClinical AIEU AI ActComplianceXAI

Selected work

Real systems. Not toy projects.

EU AI Act — Document Q&A System

A production-grade RAG pipeline that lets compliance teams query the full EU AI Act corpus and get precise, sourced answers. Hybrid retrieval with ChromaDB, structured evaluation, and a FastAPI backend built for integration.

400+ pages Corpus
Hybrid retrieval Architecture
Eval pipeline Quality assurance

Multi-Agent Research Pipeline

An autonomous research system that decomposes complex questions, routes them to specialised sub-agents, and synthesises findings with confidence weighting. Built with LangGraph for conditional orchestration and structured state management.

LangGraph Orchestration
Conditional routing Architecture
Confidence scoring Quality

About

Hi, I'm Kim.

Kim — Signal Syntax

Signal Syntax is a one-person AI consultancy built on a decade of engineering experience across hardware, research, and machine learning.

I hold a PhD in Medical Technology from NTNU, where I specialised in explainable AI for clinical applications — developing interpretability techniques for early cerebral palsy detection in collaboration with St. Olavs Hospital.

Before moving into AI, I spent 5+ years as a hardware and electrical engineer across Taiwan, the UK, Spain, Greece, and Norway. That cross-disciplinary background means I think about AI systems from the ground up — not just the model, but the infrastructure, the edge cases, and the humans who depend on it.

Based in Norway. Available worldwide.

Education
PhD, Medical Technology — NTNU
Research
Explainable AI for Clinical Applications
Engineering
5+ Years — Hardware, PCB & Power
Location
Oslo, Norway — Global Clients

Get in touch

Let's talk about your project.

Whether you need a quick feasibility check or a full build-out, we'd love to hear from you. Book a free 30-minute discovery call or send us a message.

We typically respond within 24 hours.

Prefer email? Click to reveal address