Artificial intelligence
Evaluation criteria defined before training, data provenance tracked, results recorded for reproduction.
Research
Research that moves into practice.
We conduct applied research in scientific computing, artificial intelligence and cybersecurity. We believe that research which cannot be checked by others has limited value, and we design our work accordingly.
Reproducible
Focus areas
Three areas, one standard of evidence.
Each project starts from an open question that matters to practitioners and ends in a tested system or a documented method.
Cybersecurity
Threat analysis, hardening and defensive techniques evaluated against realistic conditions.
Scientific computing
Reliable workflows on shared and high-performance infrastructure, so long calculations can be repeated and verified.
Method
From question to evidence.
The same path for every study, so results can be reviewed by people who were not in the room.
- Questionstated precisely
- Methodwritten first
- Experimenttracked and logged
- Reproduceby others
- Sharewhere appropriate
Collaboration
Work with us on hard technical problems.
We welcome collaboration with academic institutions, public bodies and industrial partners on projects that require rigorous technical work, clear evaluation criteria and sustained engineering effort.
Joint studies with shared evaluation criteria
Reproducible methods you can adopt
Engineering capacity to build what the research proves
Publication where it is appropriate
# how a collaboration starts
$virshark scope --question "..."
shared question and evaluation criteria
roles, data and publication agreed in writing
$virshark start
first milestone scheduled
Tell us what you need to get right.
Describe the context, the constraints and the outcome you need. A member of our team will respond within one business day to arrange a first discussion.