hackAtech, a new format
What is the hackatech in short? “It’s a 54-hour innovation marathon, during which participants work in teams around a start-up project, an idea using digital technology stemming from the INRIA laboratories and Irisa”, says event coordinator Antoine Le Graët.“Topics range from artificial intelligence to virtual reality, data science or cyber-security. The participants were able to draw on this catalogue of software and algorithms, in addition to the expertise of our researchers, to draft a proof of concept aimed at providing a solution to a problem or an identified need in a given market. Each team then had to present its project before a jury.”
A mix of profiles
Ten teams were formed around a project combining both scientific and entrepreneurial expertise. “Among the sixty or so participants, we were lucky enough to welcome the Master’s students in Innovation and Entrepreneurship from INSA-Rennes and the Rennes School of Business. This programme offers engineering students the possibility to follow a business management course alongside their engineering syllabus. The programme is headed by Éric Anquetil, a member of the Intuidoc research team, and Thierry Bossee Pilon, a lecturer at RSB. The hackAtech formed part of their teaching programme for the start of the academic year. They brought a great deal of enthusiasm and momentum to the event. Participants also included researchers, engineers and PhD students, many of whom, but not all, come from our laboratories.” Throughout the event, participants were assisted by the engineers from INRIA’s Experimentation and Development for Research Department (SED). For the market and business aspect, the teams were helped by experts from the support structures in the ecosystem, i.e., Poool, Village by CA, 1kubator, Pépite Bretagne and the IMT Atlantique incubator.
Four awards given
Industry-based case studies
The hackAtech is also open to representatives of the industrial sector, who come to submit concrete technological issues for which they are seeking a solution. “For this first edition in Rennes, the regional newspaper Ouest France provided three case studies.” Just the ticket! “Bertrand Coüasnon, a researcher from the Intuidoc team, had a turnkey software programme up his sleeve which provided the perfect solution to one of the problems put forward. Discussions began as of the following week with the R&D department of the company in question. This is a really positive point for us, as the aim of the event is to create a link between us and industrial companies.”
Things do not necessarily stop there for the winners of this technological marathon. “For those who are interested, we will offer to explore the subject further, both from an R&D and a market perspective, with the possible integration of the Inria Sartup Studio programme, the INRIA structure devoted to the maturation of technological projects.” This incubator is currently gaining ground. It will soon be in a position to support up to 100 projects annually on nationwide level.
Aftermovie hackAtech in Rennes (in French)
Immediately following this first hackAtech in Rennes, the idea of a second edition was raised. “This wasn’t our initial intention,” Antoine Le Graët admits. “Preparation of the event was very time-consuming. We weren’t necessarily planning on setting up a second event straight away, but the results and the enthusiasm it generated have led us to reconsider. If we do organise a second edition, it may not be led solely by INRIA. The Institution could invite other players from the eco-system to organise the event together.”
Genomic calculation at ten times the speed
Among the researchers involved in the hackAtech was Dominique Lavenier, a member of GenScale, an INRIA team specialised in bio-computer science. He has a prototype which speeds up genomic calculation, by a factor of ten!
“We’re carrying out studies with Upmen, a young company based in Grenoble which has come up with a new type of memory. In standard architecture, the processor is linked to a main memory. These are memory modules that we add to the equipment. The more memory we want, the more modules we buy.” The problem is that “at present, when applications process large volumes of data, the processor draws so much on this memory that the bandwidth between the two is completely saturated. Processors spend their time waiting for data to arrive. We come up against what we call the memory wall. We can add as many computational resources as we like, but the application won’t work any faster. To absorb this bottleneck, Upmen creates small processors directly integrated in the memory modules.”
To reap the benefits of this, however, the software needs to be adapted to the new equipment. This is where GenScale comes into play. “Our genomic applications manage high volumes of data. They are thus concerned by this bandwidth issue. We’ve been working with Upmen for 5 years on finding a way to effectively align our algorithms with this innovative memory which is similar to a parallel architecture.” The result is an in-depth rewriting which produces a sharp acceleration in processing.
Creating a start-up
There we have the theory, but there is still a step to be accomplished: “To place this technology in the hands of the end user. As things stand today, we’re missing a third player who can commercialise a solution based on the Upmen architecture and GenScale’s algorithms. Hence the idea of a creating a start-up. That’s why we decided to take part in hackAtech.”
During the event, a group of five people came together to consider the researchers’ offer. “They examined the business model to determine exactly who to propose this technology to. To bio-computer scientists? To companies already selling software in this field? Or to major pharmaceutical groups with IT departments in need of computational boosters?”
Following this first exploration of potential markets, Dominique Lavenier is hoping to take things further. “I’m hoping to get people interested in this project, to convince them to invest and see it through. Among those who took part in hackAtech, one or two people could see themselves pursuing this.” The idea is thus going forward...