Here at Corvil, generating useful insights from data has been the raison d’etre of our business for the past 17 years. We work primarily with firms who do a lot of business electronically, and who therefore depend on computers and networks to keep their own businesses running smoothly. Our role is to listen in to streaming data within these environments, and use it to generate insights that help our customers make sure their systems are optimized and secure. There’s a strong ‘real-time’ element in most of the problems we tackle, because when there’s a performance problem or a security breach in a business-critical system, our customers want to know about it immediately.
The financial sector was a strong adopter of Corvil’s products in our early years. At the time, the world of financial trading was switching over to fully electronic operation. Computers were being used not only to execute trades, but also to make the actual decisions about when, where and what to buy and sell. Making sure that these systems behave and perform as intended is obviously a priority for their owners. And the best way to do that is by continuously collecting and analysing data that reveals what’s going on, even at the inhumanly fast speeds that trading systems typically operate at.
Machine learning as a defence mechanism
In more recent years, cybersecurity has emerged as a big concern for most of our customers. Hackers have unfortunately become a lot more sophisticated, using password-phishing and zero-day exploits to break through the perimeters defences that firms relied on in the past. Once the hackers are inside, they often know exactly how to avoid triggering the static detection mechanisms that many organizations use to find them.
To cope with these threats, the industry is turning to a more dynamic type of defence that attackers will find much harder to predict. These dynamic mechanisms are often based on machine learning. The idea is to deploy ‘cybercop’ algorithms that can learn to tell the difference between a normal user and a malicious attacker, based on their behaviour. It’s a challenging problem because the attackers are simultaneously trying to disguise their activities to the extent that they can. Nevertheless, the new techniques have shown tremendous promise, and are generating a lot of excitement in an industry that’s keen to tip the scales back against the hackers.
When we founded Corvil in 2000, the term “Data Scientist” was pretty much unknown outside of research circles. I believed at the time that the growing availability of computing power and data would lead to new practical applications for maths, statistics and probability. The actual scale of the success that Data Science has now achieved, has nevertheless been surprising even to an original believer like me. The DatSci Awards in Ireland last year were a wonderful illustration of this, with high calibre contributions coming from multiple sectors of Irish industry and most of our main research institutions. I’m looking forward to participating in the awards again this year, albeit this time as a judge. Corvil will also be coming along to the awards ceremony in September to listen to the experts and enjoy the spectacle. See you there!