From Academia to Real-World Impact: Insights from Stefan Grafberger

"I first crossed paths with my future PhD advisor, Sebastian Schelter, during an internship at Amazon Research between my bachelor's and master's degrees. I was immediately fascinated by the type of work there: not only were we working on challenging research problems with exciting technology, but the work we did immediately greatly impacted customers. The ability to open-source our work, like our data quality library, Deequ, which is now widely used globally, further fueled my motivation to do impactful research."

‍

When Schelter became a professor at AIRLab, Grafberger knew it was the perfect place for his PhD, offering the opportunity to work on challenging academic problems with real-world applications.

‍

"Especially in machine learning (ML), there is usually a large gap between ML research and ML usage in industry. Existing research typically assumes clean, fully integrated datasets and expert-level proficiency in ML techniques. Meanwhile, in the real world, many practitioners may have, e.g., backgrounds in software engineering rather than ML and statistics, and often spend large portions of their time on tasks such as data loading, data cleaning, and model deployment instead."

‍

Working for AIRLab

At AIRLab, Grafberger worked on topics like data quality validation for stream processing, the validation of data preparation pipelines for ML, and data valuation for recommender systems:

‍

"Collaborating with BOL. to apply techniques such as data valuation to real-world recommender systems has been particularly exciting. Even tiny improvements in recommendation systems can have a considerable impact when serving recommendations to millions of customers.

‍

One of the highlights of my time in AIRLab was collaborating with Barrie Kersbergen, a data scientist who joined bol very early in 2010 and has developed recommendation systems for bol that serve millions of items to customers. Engaging with industry experts like Barrie and gaining insights into the practical challenges faced by companies like bol have been invaluable experiences that have enriched my research journey.

‍
Reflecting on my experiences, I am confident that choosing to pursue my PhD at AIRLab was the right decision. The opportunity to work on impactful research projects with real-world applications alongside industry experts has been fascinating. If given the chance to choose again, I would immediately start my PhD journey at AIRLab again."

‍