Research Engineer

About us

Telluvian is a London-based, VC-backed pre-seed startup. We use mechanistic interpretability, our understanding of what's happening inside neural networks, to make AI cheaper and more secure. Learn more at telluvian.ai.

The role

We're looking for a generalist ML researcher to take on the interpretability questions we're facing and turn what we learn into things we can build. Prior interpretability experience isn't required, but more general experience with ML is.

Due to the nature of interpretability, this involves a lot of very open-ended work where there is no one right answer. In some cases you may be handed a research direction, but in many cases you will be able to decide on the direction yourself.

You'll need to be a competent engineer to do this research, but only in the ways that serve it. Confident Python is the bar; frontend skills aren't expected.

You'll join a team of fewer than ten people. Your research will form a key part of our product pipeline.

This is a full time position with flexible working hours. We ask that you're in person in the London office minimum four days a week.

Start date is flexible but our target date is early September.

What you'll do

• Research interpretability questions, including choosing which ones are worth asking. You'll set your own directions as often as you're given them.
• Build prototypes from your findings. Often the end goal of a project is proving there's a tractable commercial use case for your research direction.
• Communicate clearly to technical and non-technical audiences, inside and outside the company.
• Collaborate with the rest of the team, picking up other people's directions quickly when it's useful to.
• Keep on top of published work, not easy in a field that moves faster than the literature.

What we're looking for

• A computer science or related STEM degree, with relevant internships or prior work experience.
• A drive towards understanding pre-trained models. This is the attribute we weight most heavily.
• Competence with the Python ML stack: PyTorch and Transformers primarily. You need to be a good enough engineer to run your own research end to end.
• A strong work ethic with the ability to work in a self-directed manner. You will be working with advice rather than instruction.
• Confidence and good communication skills.

Nice to have

• Production ML experience.
• Experience writing and managing custom kernels; JAX or CUDA.
• Prior interpretability work.
• A long-standing interest in machine learning and computational research.
• Commercial judgement to tell an interesting result from a useful one.

What we offer

• Real ownership of research directions from early on.
• Meaningful equity in the company.
• Flexible working and genuine research freedom.
• 28 days paid annual leave.
• Workplace pension with employer contributions.

How to apply

Send a CV and cover letter to careers@telluvian.ai. The process is an initial screening followed by two online interviews and an in-person paid work test.

We unfortunately don’t have the capacity to provide visa sponsorship at this time.

As part of your email please include your earliest available start date.