A small practice that builds production-grade AI for regulated industries — banking, fintech, legal — where reliability, privacy, and compliance are not optional.
Search that understands intent. Drafts that match your tone. Summaries that respect context. Built into your existing app, with the evals to keep them honest.
Document review, support triage, data extraction, drafting workflows. The systems your team uses every day, made dramatically faster.
For founders building an AI-native product. Architecture through launch, with the evaluation system that makes it stand up under real users.
Kubernetes, Helm, Terraform, GitOps, Azure and other clouds. Model serving, observability, cost control — proven in regulated banking environments. The plumbing that takes AI from a laptop to production, and keeps it there.
When you don't need us to build, but you need clarity on what to build. Due diligence, architecture review, fractional CTO.
A short, focused engagement to define the problem, decide whether AI is the right tool, and propose a scope. If we say no, we say no.
End-to-end, in your repo, on your infrastructure. Real data, real users, evaluated against the baseline we agreed on in week two.
Training for your team, a runbook, an eval suite, clear documentation. Then it's yours — which was always the point.
An AI-powered platform for portfolio and hedge-fund managers: risk decomposition, factor analysis, scenario simulations, stress testing. It answers one question — how does a portfolio behave when markets break?
Proprietary risk mathematics paired with conversational AI agents, so a manager can ask plain-English questions and get back rigorous analysis. Full-stack: Python, FastAPI, React, TimescaleDB, Qdrant, Celery workers, RAG pipeline — deployed on EU infrastructure.
demo.radomir.fr →Most legal-AI products cost institutional-firm prices or do shallow work. Casefile Review takes the AI tool the lawyer already uses — Claude, ChatGPT, Cowork — and connects it to their actual case files. The lawyer asks questions in plain English. The system drafts pleadings, verifies quotations character-for-character, and traces citations to bundle pages.
Validated end-to-end on a real UK Court of Appeal matter. Productisation in progress; founding cohort open to UK litigators.
The model is the easy part. The harder, more lasting work is the evaluation harness that keeps it honest after we're gone.
The person on the discovery call is the same person who will write your code, design your eval, and stand behind the result.
One number for the whole engagement. No T&M creep, no scope shuffling, no phase-two surprises.
Code that lives in your repositories, runs on your infrastructure, readable by your engineers. Nothing locked behind our tooling.
The model is the easy part. The harness that keeps it honest after we're gone is what we treat as the real deliverable.
GDPR and EU AI Act alignment from day one. RAG with permission handling, deployment on EU infrastructure — AI on sensitive data without handing it to US cloud LLMs.
About a third of intro calls end with a recommendation not to start. That's rare for a consultancy, and it can save you a quarter.
Magenta Code is the practice of Radomir Klacza — a senior platform engineer with 15+ years in banking and regulated finance: Linux at Citi, DevOps at ING, and production engineering at SIX Digital Exchange, where he ran an Ethereum staking platform on Azure Kubernetes with Terraform and GitOps.
Today he is co-founder & CTO of Fundrank, building AI risk analytics alongside Raphaël Douady, Research Professor of mathematical finance at the Sorbonne. The RAG pipelines, the eval harnesses, and the Kubernetes clusters they run on — built by the same person you talk to on the discovery call.
MIT Professional Education, Applied Data Science (2024). Published at IEEE INFOCOM. Based on the Côte d'Azur, working remotely across Europe, in English and French.
Tell us what you're trying to build. If we can help, we'll say so. If we can't, we'll point you to someone who can.