Master Thesis - Data-Driven DevOps

At Tetra Pak we commit to making food safe and available, everywhere; and we protect what's good – protecting food, protecting people, and protecting the planet. By doing so we touch millions of people's lives every day. And we need people like you to make it happen. We empower you to reach your potential with opportunities to make an impact to be proud of – for food, people and the planet.

About this opportunity
The DevOps methodology has transformed software development by integrating development and operations into a shared, automated, and iterative process. Its primary success lies in breaking down organisational silos, fostering collaboration, and promoting a mindset of continuous improvement. Through practices such as continuous integration (CI), continuous deployment (CD), and automated testing, DevOps has enhanced software delivery and reliability. However, while automation has streamlined many tasks, there remains a potential to further optimise DevOps practices using data-driven techniques.

In a data-driven culture, decisions are made based on data, given that everyone has access to the right data and the tools to interpret it. Despite the widespread adoption of DevOps, performance evaluation in DevOps implementation remains a challenge. Traditional methods, like maturity models and surveys, are often subjective and do not give teams the real-time, actionable insights they need. In this thesis, we aim to explore how a data-driven approach can take DevOps to the next level, shifting from a reactive approach to one that is more predictive and adaptive. The focus will be on developing a framework that leverages DevOps-related metrics data, advanced analytics, and Machine Learning (ML) to detect pattern, pinpoint bottlenecks, and drive targeted enhancements across the DevOps lifecycle. Data-driven DevOps does not only provide quantitative data to support improvements, but also nurtures a culture of continual improvement and empowers teams to base their decisions on empirical evidence rather than gut instincts.

This master’s thesis is part of the NextG2Com competence centre and is to be performed in collaboration between Tetra Pak and the Faculty of Engineering at Lund University.

Research Questions
This master’s thesis is supposed to answer the following research questions (RQ) in the context of Tetra Pak:
RQ1: What are the metrics data that should be collected?
RQ2: How can operational data be effectively collected and processed to provide actionable insights for DevOps improvements?
RQ3: What machine learning techniques are most suitable to predict, for instance, system failures or performance bottlenecks in CI pipelines?
RQ4: How can real-time data analytics be integrated into existing DevOps practices to automate decision-making processes such as scaling and performance optimisation?

What you will do

• This thesis will combine theoretical research, practical development, and possibly evaluation, following these key phases:
• Literature Review: Conduct a review of the state of the art on DevOps performance evaluation with a focus on data-driven approaches.
• Metrics Selection: Select relevant and key metrics, design a data collection architecture, and set up a tool to collect data automatically.
• Data Collection and Analysis: Develop a framework to collect and visualise data from DevOps tools like Azure DevOps. The data may include logs, error rates, and resource utilisation metrics.
• Case Study: Select a suitable case study and apply the framework to.
• ML Model Development and Integration: Use and, if feasible, integrate ML models into, for instance, a live CI pipeline to identify patterns and predict potential failures.
• Evaluation and Benchmarking: Aim to conduct initial testing of the data-driven DevOps pipeline against a traditional pipeline.

Related work
Recent research shows an increasing focus on automating the process of collecting and analysing DevOps metrics, such as the following two theses:

• Automatisering av DORA-mätningar inom DevOps
• DevOps Approach to Measure Product Quality and Developers Efficiency

We believe you are
Currently studying in the final year of a master’s degree in computer science, software engineering, or a related field.
Knowledgeable in machine learning, with an interest in learning about DevOps practices as a plus.

Apply Now!
Please submit your CV and brief motivation letter describing your interest in this area.
This posting expires on 2024-11-17.
If you have questions about this opportunity, please contact Al-Hussein Hameed Jasim at alhusseinhameed.jasim@tetrapak.com.

Diversity, equity, and inclusion is an everyday part of how we work. We give people a place to belong and support to thrive, an environment where everyone can be comfortable being themselves and has equal opportunities to grow and succeed. We embrace difference, celebrate people for who they are, and for the diversity they bring that helps us better understand and connect with our customers and communities worldwide.