Data Analyst & Data Scientist · PhD Physicist · turning complex data into decision-ready insights

📍 Darmstadt, Germany

For 8+ years I analysed large, complex datasets in particle physics, first in my PhD, then as a postdoc at Goethe University Frankfurt working with GSI/FAIR. My job was to separate real signals from noise and to make sure the data could be trusted before anyone drew conclusions from it.

Now I bring that same rigour to business questions. I'm completing the Data Science Bootcamp at WBS CODING SCHOOL, where I turn raw data into clear recommendations with SQL, Python, Machine Learning and Tableau.

• 🔬 Analytical rigour - statistics and machine learning to find what's real in noisy data
• ✅ Data quality - I built an automated QA framework that caught data errors before analysis
• 🗣️ Communication - 20+ peer-reviewed publications explaining complex results clearly

A/B Test of Homepage Buttons · Python · SciPy

Compared four button designs across ~100,000 homepage visits with a chi-square test and Bonferroni-corrected comparisons.

Result: white buttons roughly doubled the click-through rate of red ones (2.0–2.1% vs. 0.8–1.1%).

Discount Strategy Analysis · Python · Pandas

Cleaned three raw sales tables (8,700+ duplicate rows, malformed prices) to test whether discounts pay off.

Result: sales stop rising above a 20% discount → recommended capping discounts and shifting them to high-end products.

Market-Entry Analysis · MySQL · Tableau

Queried a 9-table marketplace database to decide whether an online tech retailer should partner with a Brazilian platform.

Result: ~93% of orders arrive on time and tech makes up ~14% of revenue → recommended signing the deal.

Smartwatch Activity Dashboard · BigQuery · Tableau

Cleaned 960 days of my own Samsung Health data and built an interactive dashboard.

Result: my 10,000-step goal rate rose from under 30% to 73% in 2026.

Playlist Clustering for a Music Startup · Python · scikit-learn · Spotify API

Grouped 5,000+ Spotify songs into 34 playlists (50–250 songs each) with PowerTransformer, PCA and K-Means, then posted them to Spotify.

Result: good first drafts in seconds, but an expert is still needed: the audio features capture a song's mood, not its genre.

Cloud ETL Pipeline · Python · Pandas · MySQL · Cloud SQL · Cloud Functions · Cloud Scheduler

Automated pipeline that scrapes city data from Wikipedia, pulls weather and flight data from two APIs, and loads it into MySQL on Google Cloud on a schedule.

📝 Read what broke and how I fixed it: Scraping, APIs and Cloud SQL: What My First Data Pipeline Taught Me

QA Data Framework · Python · C++ · CI/CD

Framework that automatically checks large scientific datasets for errors, with quality checks built into the CI pipeline.

Thanks for stopping by!