Applied software analytics to help managers on their DevOps transformation

A large number of companies look for DevOps transformation, so they are more agile to be competitive in the market. The practices that are related to the DevOps is what allows companies such as IMVU to release as fast as 50 releases/day. They are the same practices that allow large companies such as Google and Netflix to be fastly evolving with a few quality issues. However, making a DevOps transformation is not a straightforward exercise. Such a transformation is full of uncertainties and challenges, as it requires a good comprehension of the existing software development and releasing process, defining the appropriate strategies for a DevOps transformation, and measuring the impact of a DevOps transformation.
As the DevOps transformation is not straightforward, the goal of this project is to equip managers of our industrial partners with the appropriate solutions so they can better manage their transformation. The goal of this PhD project is to investigate, identify, and develop mechanisms to help the managers of our industrial partners improve their DevOps process. To do so, this project will consist of collecting a set of metrics to first understand the current process of our partners, as well as compare it to a set of open source projects. Explore the potential of leveraging statistical and machine learning approaches to help our industrial partner make better decisions (e.g., identify the optimal size of a code change) to improve its DevOps process. Explore different approaches that are related to the deployment of these statistical machine learning approaches in an industrial context, as well as their limitations.
This project will cover existing and recent topics such as: DevOps, release engineering, statistical analysis, machine learning, mining software repositories, ….
 

Connaissances requises

We are looking for students with good programming skills (e.g., python, R, Java), good knowledge about the software engineering activities (e.g., testing, integration, configuration, ...), and some basic knowledge on machine learning (e.g., logistic regression, random forest, deep learning, ...).
 

Programme d'études visé

Doctorat

Domaines de recherche

Les systèmes logiciels, le multimédia et la cybersécurité

Financement

Bourse disponible /Funding available

Autres informations

Date de début : Automne 2022/ Fall 2022

Autre personne contact :  Professeur Mohammed Sayagh , mohammed.sayagh@etsmtl.ca