Work on hard problems with a small team

Sedai is a quantitative trading company in Luxembourg. We build our entire stack in-house — market data, research, model training and live execution — and engineers here own problems end to end.

What we look for

Strong programming, and taste in code

We write code a teammate can pick up cold: readable and testable over clever. We care more about how you think than which language you have lived in.

Systems thinking under ambiguity

You can design something you will still be extending next year: cleanly separated, with your assumptions written down and room to grow. And you keep moving with incomplete information instead of waiting for a perfect spec.

A feel for failure modes

Our systems meet malformed feeds, partial fills and adversarial inputs, where being correct and being fast both matter. You instinctively think past the happy path.

Quantitative rigor

Probability and statistics, empirical skepticism. We test our own ideas against the data before using them.

Ownership

You see what needs doing and do it, without waiting to be asked.

Direct communication

Knowledge is shared by default. Ideas get argued hard; people don't.

Culture

Small by design

Flat, and everyone sees the whole picture — from raw market data to live P&L. People learn by working next to each other, not in one-off training sessions. There are no layers between a question and the person who can answer it.

Lunch, and the game after lunch

We eat together, and then the cards come out — Good Cop Bad Cop, Vampire Queen, and whatever someone brings back from a trip. Puzzles and board games make regular appearances too. It is not a team-building exercise; it is just what lunch looks like here.

Arguments are about ideas

Decisions get made by argument, and the best argument wins, regardless of who makes it. Debate is direct and goes after the idea, never the person. Successes and failures get talked about quickly.

The office

Our office is in Luxembourg City.

Where you might fit

Software Engineer

The trading platform itself: services that ingest market data, evaluate models and send orders — all in-house, on a small set of well-understood pieces.

Data Engineer

The path from raw market feeds to clean, adjusted, queryable history — pipelines that never lose a tick and never rewrite the past.

Quantitative Researcher / ML Engineer

Feature computation, model training and evaluation — the machinery that turns a hypothesis into something that can trade.

The stack is Java, gRPC, Kafka and Cassandra — all built and run in-house, so a small team owns the whole path from a raw market tick to an executed order. At the busiest moments of the trading day, that path carries tens of thousands of market events per second.

We expose our services to AI coding agents through MCP, so agents do bug hunting, run experiments, evaluate results and extend existing research ideas. Machine learning drives the models themselves, and trading is fully automated — from signal to executed order.

Your code runs against live markets within weeks, not quarters, and the people who built the systems are the people you sit next to.

How to apply

Send an email to contact@sedai.lu — a short note about what you have built and what you want to work on. The process is an intro conversation followed by technical conversations with the team, about real problems rather than puzzles. We work from our office in Luxembourg City three days a week, and from home the other two.