01Careers
Growing, carefully, and saying so honestly.
02Open positions
No listed vacancies at this moment.
We are not going to post roles that are not funded, or invent a hiring pipeline to look larger than we are. When a position opens it will appear here with a real scope, a real range and a real start date.
That said — a studio this small hires from conversations far more often than from adverts. If the areas below describe you, write. A specific email about a specific problem is read carefully, and answered.
lluyotlabs@gmail.com03Future opportunities
The four kinds of people this studio needs.
These are the shapes of the roles as they will exist, described so you can tell whether the work would suit you before either of us spends an afternoon on it.
Applied Mathematics & Modelling
You formulate. Given a vague problem and a domain you do not yet know, you can produce a model with stated assumptions, a solution method, and an honest account of where it stops being valid.
- Modelling
- Optimization
- Numerical methods
- Statistics
Machine Learning & AI
You build systems around models rather than demos of them: retrieval, evaluation, calibration, failure modes. You are more interested in what a model gets wrong than in what it gets right.
- ML engineering
- LLM applications
- Evaluation
- Python
Product Engineering
You ship. Web, mobile or both, with a real concern for the second year of the codebase's life — and enough design sense to know when an interface is lying about how confident the system is.
- TypeScript
- React / Next.js
- Mobile
- APIs
Data Science
You decide whether a result means anything. Experimental design, causal reasoning, and the discipline to report the confidence interval that makes the story less exciting.
- Inference
- Experiment design
- Causal analysis
- Visualisation
04What it is like
Four things to know before you write.
Two of these are advantages and two are costs. Which is which depends entirely on what you want from the next few years.
Small by design
Lluyot Labs is a young, independent studio. That is stated plainly because it is the most important thing a candidate needs to know: there is no large team behind this, and the work is correspondingly broad.
Depth over breadth of tools
We would rather understand one method properly than integrate five services. Expect to read papers, derive things, and defend a formulation before writing the code.
Own a problem end to end
The same person who formulates a model usually ships the product around it. It is demanding, and it is the reason the results are coherent.
Remote and asynchronous
Work happens in writing. A clear document beats a meeting, and a reproducible notebook beats a confident opinion.
05Applying
What actually helps us decide.
- 01
One problem you solved properly
Not your whole history. One piece of work, what made it hard, what you tried that failed, and how you knew the final answer was right.
- 02
Something we can read
A repository, a paper, a write-up, a shipped product. Anything where we can see your reasoning rather than a claim about it.
- 03
Why this studio
Two sentences. If they would apply unchanged to fifty other companies, they will not tell us anything.
Every message is read by someone who does the work. Replies take days, not weeks — and a no is still a reply.