Mathew Lodge

Mathew has over twenty-five years of experience in the software industry in developer, product and marketing roles. Before joining Diffblue, his titles included SVP at Anaconda and Vice President of Cloud Services at VMWare. In each role, his focus has been on building and marketing products that customers love.


Using Reinforcement Learning AI to accelerate DevOps

You probably learned that hacking on code until it works is the wrong way to write a program. But today, that’s essentially what we do in AI (albeit trillions of times faster)—and it achieves remarkable results!

Reinforcement learning was notably used by Google’s AlphaGo algorithm to teach itself how to beat human Go grandmasters. This method is especially useful when the search space of solutions is polynomial: at some stages in a game of Go, there are more possible moves than atoms in the universe. Instead of brute force, the algorithm uses reinforcement learning to do probabilistic search of potential solutions and make the move that’s most likely to result in a win.

In this session, hear about how we apply this technique (along with others) at Diffblue to generate unit test programs for Java projects, hacking the code using reinforcement learning to find the tests that achieve coverage and test usefulness goals. The resulting test suites find more problems earlier in DevOps pipelines, helping companies like Goldman Sachs adopt and maintain rapid, high-quality code delivery.