Scientific computing for companies where the science and the engineering have to understand each other.
I'm Tjelvar Olsson. I work with companies whose core business is science — helping them build computing systems that scale, manage their data properly, and make sure critical technical knowledge doesn't live in just one person's head.
Why the intersection matters
The value is in the combination. Without understanding the science, it's hard to know what shape the computing needs to take. Without understanding the computing, it's hard to know how analysis can be restructured to run more efficiently. Most specialists are strong on one side. I work at both.
Computing
HPC platforms, workflow orchestration, cloud pipelines. Production-grade from the start, not just working on your machine.
Science
Biochemistry PhD from UCL. I read the methods section, not just the spec.
People
Every engagement is built around knowledge transfer and documentation — so systems run without me once the work is done.
Three areas where I have a track record
I've done this work from the inside of scientific organisations — at research institutes, in industry and now independently.
01
Computing that scales with the science
Your setup made sense when you started. Since then the data has grown, the team has changed, and what worked at the beginning is starting to creak. Jobs take longer than they should. Pipelines break when something changes. The knowledge of how it all fits together lives in one or two people's heads.
I come in, learn the system and the science behind it, and help you build something that scales. That sometimes means simplifying. It sometimes means accepting that scientific computing genuinely needs a certain amount of complexity — but it should be the right complexity, not the kind that's just accumulated over time.
This is embedded work. I'm with you for months, not days. When a project calls for expertise beyond mine, I bring in specialists from my network.
02
Data that makes sense
Scientific data is only useful if you can find it, trust it, and hand it to someone else without a long explanation of the folder structure.
I built dtool — an open-source tool for managing scientific datasets. It's been adopted by a group in Germany, and one of their researchers is now introducing it to a group in Japan. Small, but real. Data management is something I've thought about seriously for a long time.
I've also spent time as the person sitting between research teams and central computing infrastructure — writing the scientific case for hardware investment, translating what the science needs into something the engineers can act on, and translating what the engineering can do back to the scientists.
03
From raw data to answers
You have image or video data — cells, organisms, agricultural samples, materials — and you need to know what's in it. The question is usually specific. The challenge is usually the pipeline: getting from raw data to a reproducible answer that doesn't require starting from scratch every time.
I scope these as fixed projects. A short feasibility phase first, so you know what you're getting into before committing. Two days of knowledge transfer included, so the capability stays with you when the project ends.
Most projects fall between £4,000 and £8,000.
How I work
I'm a consultant, not an employee. My contracts are scoped accordingly.
Fixed scope or embedded
For data analysis work, I work to a fixed scope with a defined output. For infrastructure work, I work embedded — usually over several months. The contract shape follows the work, not the other way around.
Specialist network
I don't know everything. When a project calls for expertise I don't have, I bring in people from my network who go deeper in the areas that matter. You'll know when I'm doing that.
Knowledge transfer as standard
The goal, especially for embedded work, is to leave things in better shape than I found them — properly documented, with onboarding that makes sense to the next person, and systems that don't depend on me to keep running.
Systems that don't depend on me
The work should stand up when I'm not around. That means documentation, knowledge transfer, and systems whose logic doesn't live only in my head.
I've worked with some good people over the years. Here's what a few of them said.
Tjelvar understood our biology as well as the software. He built analysis tools we can run and maintain ourselves. Everything was scripted, reproducible, and properly documented — that matters when the team needs to keep using it without him in the room.
Tjelvar has experience in working with people from a variety of backgrounds, which helps him to quickly understand the core data management requirements when approaching him. I recommend him very highly.
Tjelvar is friendly, very capable and dependable. He has an excellent understanding of the demands of biologists and has helped me and my lab members on numerous occasions with finding creative solutions to safely store and retrieve our data cost-efficiently.
Tjelvar Olsson, PhD
Scientific computing — bridging science and engineering
I have a PhD in biochemistry from UCL, and I've spent the years since moving toward the computing side of scientific work — not because the science stopped mattering, but because the computing is where a lot of scientific work gets stuck.
The combination turns out to be useful. Without understanding the science, it's hard to know what shape the computing needs to take. Without understanding the computing, it's hard to know how analysis can be restructured to run more efficiently.
I built dtool, a data management tool for scientific datasets — adopted by a group in Germany, and one of their researchers is now introducing it to a group in Japan. Small, but real. I wrote A Biologist's Guide to Computing — freely available, and found by more than 400 people through the mailing list alone, with many more arriving via links I've long since lost track of. Some of the code has aged badly. The chapters on how to think computationally haven't.
I spent time at the John Innes Centre as the interface between the research teams and the central computing infrastructure — the person who could explain what the science needed to the engineers, and explain what the engineering could offer back to the scientists. That's more or less what I still do, just independently.
I work with a small number of clients at a time. If your company's core work is science and your computing needs to grow up, I'd like to hear about it.
Research from the inside
I've been a researcher, an author, and an engineer. That breadth is the product, not an accident.
Published researcher
Peer-reviewed publications spanning computational biology and scientific software. I understand what reproducibility means in practice because I've been on both sides of a failed replication.
Google Scholar profile →Author
I wrote A Biologist's Guide to Computing — available free online. Written to be genuinely readable for biologists: the goal was always that it would be picked up and finished, not just bookmarked.
Read the book →Knowledge transfer
Knowledge transfer is built into every project I deliver — documentation, onboarding, and time spent making sure the work sticks once I'm gone. I've also run training courses in Python, Linux, and HPC for researchers at the John Innes Centre. Getting the work to stick is part of the work.
EU Horizon participant
Named as data management partner on an active EU Horizon project. I understand the data management and governance expectations that come with European research funding.
Get in touch
The best starting point is usually a short conversation.