Name: Kashf Ali
Class Year: 2028
Majors: Computer Science & Mathematics

Internship Organization:  University of Freiburg
Internship Title: Research Assistant — NLP & Media Analysis
Location: Freiburg, Germany

What’s happening at your internship? We would love to hear what kind of work you are doing!

This summer I’m working as a research assistant at a psychology lab in Freiburg, Germany, on a project that sits right at the crossroads of computer science and psychology. The goal is to understand how German-language news media covers the United Nations’ 17 Sustainable Development Goals, things like gender equality, clean water, climate action, and poverty reduction, and whether there are goals that barely get covered at all.

My job is to build the data pipeline that answers that question. I’m working with a dataset of millions of German news sentences since 1995, running each one through a multilingual AI model that reads the sentence and assigns it to whichever UN goal it’s most about. I then analyze the results to find patterns; which goals dominate the headlines, which ones are essentially invisible, and whether different outlets cover different goals. So far, gender equality is the most covered goal while no poverty and decent work barely register. Those gaps are exactly what the research is trying to explain.

Why did you apply for this internship?

I’ve always wanted to find a version of computer science that felt connected to something bigger than just the code itself. I’m drawn to questions about people; how they think, what they pay attention to, what shapes their understanding of the world. When I came across this opportunity to combine NLP with psychology research on something as meaningful as global sustainability goals, it felt like exactly the kind of interdisciplinary work I’d been hoping to find. I also wanted real research experience like something where the decisions I made actually affected the findings. This internship gave me both.

What has been your favorite part of this internship?

The moment when things finally clicked.

I spent a few days early on trying to find an AI model that could actually read German text well enough to be useful for the project. First one gave completely wrong results. Second one crashed on loading. By the third attempt I was pretty much just hoping for the best and then it actually worked. A sentence about climate change and oceans came back labeled correctly. A sentence about a gender equality law hit 96% confidence. It was one of those quiet moments where you just sit back and think, okay, we’re doing this.

What I didn’t expect was how interesting the results themselves would turn out to be. After running the pipeline on thousands of German news sentences, patterns started emerging: “Gender Equality” was everywhere, showing up in nearly 17% of matched sentences. “No Poverty” barely registered at 0.1%. That’s not just a fun data point, it’s actually saying something real about which global issues German media thinks are worth covering and which ones it essentially ignores. The research question suddenly felt very alive in a way it hadn’t when I was just building the pipeline.

But honestly, it’s not just the work. Freiburg itself has been a huge part of what makes this summer special. It’s a small city in the southwest corner of Germany, right at the edge of the Black Forest, and it has this energy that’s hard to describe — sunny, relaxed, full of students, with little streams called Bächle running along the streets that you will absolutely step in at least once. My colleagues have made it even better. Lunch breaks turn into long conversations, someone always knows a good spot to grab food, and the whole environment just makes you want to show up every day.

That combination — work that actually means something, in a place this good, with people this easy to be around — has made this a summer I’ll genuinely remember.

Was this internship what you expected it to be?

Honestly, not at all — but in the best way.

I expected something more straightforward. Build a pipeline, process some data, hand it off. What I didn’t expect was how many judgment calls were actually mine to make. Which model to use. How confident the AI needs to be before we trust its answer. How to handle sentences that are technically ambiguous. None of that has a right answer you can just look up; you think it through, discuss it with your supervisor, and make a call. That felt new and a little uncomfortable at first, and then really exciting.

I also didn’t expect how much the environment itself would matter. The colleagues who pull you into lunch conversations, the supervisor who asks what you think instead of just telling you what to do and it all adds up. I came here to learn how research works. I didn’t realize I’d also get to feel like an actual part of it.