There isn't really a single Rylivo type. The people who end up here come from biology, from chemistry, from machine learning, from software engineering, from clinical medicine, from physics, from places it's hard to describe in a single discipline. What they tend to share is harder to put on a CV: a long-running, slightly stubborn interest in two fields at once, and a visible impatience with the gap between what science can do and what patients actually receive.
The team is small enough that you'll meet most of it in your first week and recognise the rest by the second. It is also varied enough that, on any given day, the most interesting conversation in the office might be about molecular dynamics, training-set curation, manufacturing logistics, or how to write a better post-mortem. Nobody owns all of that. The point of the team is that, between us, we don't have to.
Who joins us
Some of us spent years in academic labs and decided that the bottleneck on their work was no longer scientific imagination but engineering scale. Others came from large industry research groups and wanted a place where a clean idea could become a running system in a week instead of a quarter. A meaningful portion came from the open-source AI world — comfortable with very large models, very large codebases, and the discipline of writing software for other people to read. A surprising number have published in both biology and machine learning, which we mostly take as evidence that they couldn't make their interests behave.
We think this mix matters. Drug discovery has, for a long time, organised itself around fenced-off specialisms: chemistry over here, biology over there, modelling at the edge, the clinic far away. The questions we want to ask don't respect those fences. So we've built the team to ignore them too.
What we look for
Taste, before anything else. By taste we mean a sensitivity to which experiments are worth running, which models are worth training, which ideas are worth chasing for six months and which are worth sketching out in a paragraph and shelving. Most of the work in research is choosing what to do next. We try to hire people who choose well.
After that: rigour, in the practical sense — the willingness to run the extra control, write the careful evaluation, read the paper that almost certainly contradicts the conclusion you'd like to reach. Then patience — the kind that comes from realising you're working on a problem whose timescale is human lifespans. And then, finally, generosity, because the kind of cross-disciplinary work we want to do only happens when people are willing to teach each other the boring parts of their fields.
How it feels
Most days are a mix of long quiet stretches of focused work and short, animated discussions in front of a whiteboard or in a chat thread. We spend a lot of time reading — papers, internal memos, each other's pull requests. We try to write more than we talk; the company has a memo culture for the same reasons every other thoughtful organisation does. Decisions live in documents you can search later, not in the corners of meetings you forgot to attend.
The pace is brisk but not breathless. We are aware that the people whose lives our work might one day affect are waiting; we are also aware that good science cannot be hurried into being. The discipline is to optimise for the things that compound — model quality, data infrastructure, lab throughput — and to leave room for the things that don't.
What this isn't
It is not a place to prove a single technique. It is not a place to ship one paper a year and call it a career. It is not, particularly, a place to be the smartest person in the room about your specialism — most rooms here already have one of those, and what we'd like from you is the second thing.
It is a place to take a hard problem seriously over a long horizon, with colleagues who care about getting it right and patients who deserve us to hurry. If that sounds like the kind of team you'd like to be part of, we'd love to hear from you.