Opportunities
Open problems, worked on carefully.
We add people rarely, and only when there is specific work that needs doing. When a role is open it is listed on this page, with a description of the work.
Current openings
Open roles
None at present.
There are no open roles. Introductions from people whose work bears on the problems below are still read.
What we work on
- Inference that stays calibrated when the process generating the data changes.
- Learning systems that adapt without forgetting what still holds.
- Evaluation that resists leakage, selection effects and overfitting to the past.
- Decision-making where actions change the environment being modelled.
- Infrastructure that makes research reproducible and operations auditable.
What the work asks for
- Measurement
- A preference for measuring over asserting, and the willingness to report that something did not work.
- Range
- Movement between mathematics, code and the untidy details of real data.
- Responsibility
- Concern for how a system behaves at its worst, not only at its best.
- Patience
- Willingness to build foundations whose value appears later.
Introducing yourself
If you believe your work bears on these problems, write to [email protected] with a short note: a piece of work you stand behind, and a question you are still working on. A link is better than an attachment. We read every note, though we may not be able to reply to all of them.