A 12-week placement with an industry partner was a unique and exciting element of the CRT in Foundations of Data Science that set it apart from more  conventional PhD programmes. Needless to say, I was very excited about my industry placement with Analog Devices, a global leader in the manufacture of sensors and signal processing integrated circuits. Beginning on June 5th 2024, I joined Analog Devices during a period of rapid change; several recent hires had brought the company a wealth of talent in the areas of data science, artificial intelligence and machine learning. Under the supervision of George Healy and Dennis Dempsey, I was eager to get started.

The goal of my project was to determine whether unusual device performance could be explained by random mismatch effects in the semiconductor fabrication process, or instead by systemic issues arising during the design or manufacturing stages. To that end, I collaborated with team members based in the U.S. and the Philippines to identify and minimise systemic issues, bringing device performance in line with a well-established statistical model.

Building on the work of previous CRT students, I was tasked with understanding how and why our exploratory data analysis did not match the industry standard model. Through design of experiments and by carefully examining the data, we found that the existing model would need to be modified to reduce the mismatch we saw with our test chip. Specifically, we identified two dominant factors which were contributing to the problem but not included in the model. These findings will be used to refine the previous model, helping Analog Devices design teams to optimise their chip designs.

Before doing any data analysis, however, I spent my first few weeks reading about semiconductor mismatch and failure modes while also drawing on the expertise of more senior team members. Then, using this new knowledge, I gathered and reformatted code written by previous interns and co-op students. My objective was to create a bespoke Python package that could easily read and analyse the data, making the previous code more reusable and enabling the team to quickly glean key insights from new experiments. I also created an updated set of instructions on how to configure the Python environment, reducing the onboarding required for any new team members.

Overall, contributing to an industrial research project has been stimulating, challenging and rewarding. Gaining the ability to quickly make meaningful contributions to a project, despite only picking up the basics a week or two earlier has been an incredibly satisfying experience. Furthermore, I was impressed by the level of ingenuity required to find applications for basic research that are novel, profitable and aligned with a company’s interests. I am certainly glad to have had the opportunity to develop the kind of creative thinking required in a fast-paced industrial setting and hope to continually improve this skill throughout my career as a researcher.

On the whole, I thoroughly enjoyed my industry placement and found it to be fulfilling, both personally and professionally. I am confident that the professional connections made with Analog Devices will stand me in good stead for many years to come.