# Julius Uhlmann
I'm Julius, a Data Scientist drawn to LLM research, data, and any kind of deep dive.
## Experience
-
February 2026 -
July 2026
AI Engineering Intern, BMW Group
- Architecting and developing LLM-based agentic systems
-
August 2025 -
January 2026
Working Student Data Engineering, Allianz
- Implemented data pipeline, from data loading to dbt transformations
- Assisted with the development of Python-based data products
-
March 2025 -
July 2025
Data Engineering Intern, Allianz
- Developed a central Python module for Snowflake data pipeline, supporting eventual deployment to production
- Designed and implemented data quality monitoring framework
-
March 2024 -
June 2024
Lifetime Forecasting Project, Süddeutsche Zeitung
- Developed customer lifetime forecasting models (scikit-learn, tensorflow)
- Applied statistical methods, tree-based approaches and basic neural networks
- Collaborated in a team of four; achieved an individual grade of 1.0
-
August 2023 -
May 2024
Research Assistant, Technical University of Applied Sciences Augsburg
- Processed data for a virtual reality study, including cleaning, EDA, and implementing physiological synchrony metrics in R
## Education
-
October 2022 -
February 2026
Data Science B.Sc., Technical University of Applied Sciences Augsburg
- Overall Grade: 1.2
- Thesis: "Assessing RLVR's Efficacy in Solving Previously Intractable Problems with LLMs" available under https://doi.org/10.60524/opus-3084
- Tutor for Analysis I and II
-
August 2024 -
December 2024
Exchange Semester, University of Oklahoma
- Focused on C++ & CS Fundamentals
- Unanimously elected by classmates to represent Public Speaking class in Josh Lee Speech Competition (JLSC)
## Skills & Interests
- Languages
- German (native), English (C1)
- Technology
- Python (pytorch, sklearn, dbt, pandas, ...), R, C++, SQL, AWS (Glue), Snowflake
- Interests
- LLM Research, Handbalancing, Tricking
- Things I Like
- Developing intuitive understandings of how things work, Reinforcement learning from verifiable rewards, Optimizing daily life, Gradient-boosted trees, To-do apps where tasks can be nested hierarchically