A summary of my academic history.
| Degree | Institution | Specialization | Completion | GPA |
|---|---|---|---|---|
| B.S. in Computer Science | DHBW Stuttgart | Computational Data Science | 2026 (Expected) | 1.6 |
| Allgemeine Hochschulreife | Gymnasium Ebingen | - | 2023 | 1.1 |
Key professional roles, including the internships completed as part of my corporate Bachelor’s degree.
| Duration | Organization | Role | Achievements and Technology Stack |
|---|---|---|---|
| since 09/2025 | DHBW Engineering e.V. | Formula Student – Autonomous Driving | Developed and optimized driving controllers for the eSleek 2026 autonomous race car. Implemented various lateral and longitudinal controllers (PID, MPC, Pure Pursuit, Stanley). Performance benchmarking was conducted in both a Speedgoat-based Hardware-in-the-Loop (HiL) simulation environment and on the physical track. Technologies: C++, Control Theory |
| 06/2025–09/2025 | Mercedes-Benz AG | Internship: Token Pruning for Reasoning-VLMs | Optimized inference time of Vision-Language Models through targeted token removal. Achieved 55.9% reduction in computational operations per second (FLOPs) while maintaining model accuracy. Technologies: Python, Transformers and Vision-Language Models |
| 12/2024–03/2025 | Mercedes-Benz AG | Internship: Benchmarking RAG Methods | Evaluated various Retrieval-Augmented Generation (RAG) approaches (NaiveRAG, GraphRAG, LightRAG) against complex standards and legal documents. Assessed key metrics including inference time, answer quality, and operational costs. Technologies: Python, LLMs and RAG |
A breakdown of my programming skills.
| Language | Proficiency |
|---|---|
| Python | Advanced |
| C++ | Intermediate |
| Java | Intermediate |
My proficiency in different languages.
| Language | Proficiency |
|---|---|
| German | Native |
| English | Fluent |
| French | Basic |
A list of papers I have authored or co-authored.
Andrea Ramazzina, Tobias Haab, David Fitzek, Stefano Gasperini, Jonas Uhrig, and Mario Bijelic. (2025). "Reasoning with Fewer Eyes: Efficient Visual Token Withdrawal for Multimodal Reasoning." NeurIPS 2025 Workshop on Efficient Reasoning.