The research team behind the extremely difficult to hack microkernel seL4 received vital funding until the end of the year from the University of New South Wales. The team, known as Trustworthy Systems at CSIRO, was remarkably abandoned by the agency earlier this month as part of a restructuring that will cut 70 jobs.
New funding from the UNSW School of Computer Science and Engineering will allow most of the Trustworthy Systems team – more than a dozen at CSIRO and a similar amount at UNSW – to stay together and to grow the tech ecosystem over the next year through his independent business. Foundation seL4.
The money will allow the team to get through a “critical” period between CSIRO’s dissolution of the shock and the foundation’s scale-up, according to Dr. Gernot Heiser, UNSW science professor and seL4 developer.
“It gives us the confidence that we can keep the team together,” he told InnovationAus.
However, several members of the core team could leave due to layoffs, but Dr Heiser hopes to keep them involved in the ecosystem.
“There will be a skills drain which we are trying to do our best to remedy by getting other people to leave [the team] recently returned, etc. But in general, that puts us in a very, very good position compared to where we were a week ago.
The seL4 microkernel provides foolproof separation between software systems so that hackers cannot access critical parts of the operating system by entering through a poorly protected hatch. The Australian team behind her are considered world-class researchers.
But CSIRO’s Data61 business unit pulled funds from Trustworthy Systems projects earlier this month as part of a new strategy more focused on artificial intelligence. The agency said the technology was “mature” and “well supported outside the organization.”
Dr Heiser disputed this claim, however, saying that there is still a lot of research to be done on the technology and the ecosystem to support it.
“The point is, although seL4 is mature enough to be deployed in the real world, there is still a lot of basic research work on seL4 itself, and there is still a lot more research on how to make computer systems. reliable in the real world, ”he said. written on his personal blog Monday.
“It’s not that just sprinkling a little SEL4 Fairy Powder on a system will make it trustworthy.”
Several international buyers quickly lined up, hoping to bring in the research team in-house, including a Chinese automaker and Singapore’s national R&D agency. InnovationAus understands that there was at least one other large company that expressed interest in taking over the team.
But funding through UNSW will keep open-source technology more widely deployed and enable the completion of several ongoing seL4 projects.
“At the moment, the [seL4] The skill set is only for the Trustworthy Systems team, ”Dr Heiser told InnovationAus.
“It would have been a real disaster if the team had dispersed. Many of these projects that are underway or about to materialize cannot materialize without this kind of critical mass of skills.
“And I still think someone would destroy that.”
The changes mean that there will be few SEL4 skills left at CSIRO, which has yet to deliver several multi-million dollar contracts involving the technology.
A spokesperson for CSIRO told InnovationAus last week that the agency “will strive to minimize impacts on partners or stakeholders” resulting from the changes.
As part of a new focus on artificial intelligence and other areas, CSIRO’s Data61 will downsize other sections of its workforce, with up to 70 jobs and seven research programs to risk.
While the agency said the workforce will eventually return to current levels, it expects this to take up to two years.
The CSIRO Staff Association called the move a “shock” and particularly concerning given the agency’s increased funding and its importance to the federal government’s digital agenda.
Dr Heiser said that within a year he hoped to have stepped up the seL4 Foundation and attract more funds for low-profile projects.
You know more? Contact James Riley by email or Signal.