CV

PDF versions: short CV · long CV

Education

• M.S. in Machine Learning, University of Tübingen, 2026 (expected)
• M.S. in Computer Science, Politecnico di Milano, 2025 (Erasmus Exchange, February 2025 - August 2025)
• B.S. in Computer Science, University of Freiburg, 2023 (October 2019 - September 2023)
• Aeronautics and Space Engineering, University of Stuttgart, 2019 (October 2018 - September 2019)

Publications

• Robin Uhrich, Marco Bagatella, Núria Armengol Urpí, Pavel Kolev, Georg Martius. Occupancy Modeling via Drifting: One-Step Generation Enables Direct Policy Extraction. Preprint, 2026. arXiv:2610.19883

Work Experience

• November 2026 - December 2026: Visiting Researcher (upcoming)
  • University of North Carolina at Chapel Hill
  • Host: Prof. Weitong Zhang
  • Research stay on offline reinforcement learning with expressive probabilistic samplers
• October 2025 - March 2026: Student Assistant
  • University of Tübingen, Autonomous Learning Group
  • Supervisor: Prof. Georg Martius, Jens Beißwenger, Anna Manasyan
  • Duties included: Working with humanoid robots, development of VLA (Vision-Language-Action) models applied to humanoid robotics
• August 2024 - February 2025: Student Assistant
  • BG Clinic - Tübingen, Hand, plastic, reconstructive and burn surgery
  • Supervisor: Dr. Cosima Prahm, Dr. Dominik Steiner, Daniel Höglinger
  • Duties included: Developing software interfaces, implementing deep learning pipeline for EMG to pose data, implementing convolutional, recurrent, and transformer architectures for time series data, making NinaPro dataset accessible through PyTorch
• July 2024 - February 2025: Freelancer
  • Volatilis, Ludwigsburg, Germany
  • Duties included: Implementing a browser-based simulation engine for risk modeling, business and operational development
• December 2023 - May 2024: Student Assistant
  • AI-Center Tübingen, Central Office
  • Duties included: Writing a wiki containing organizational aspects for international PhD students and post-doctoral researchers, performing and analyzing a department-wide survey on experiences of international researchers
• April 2023 - August 2023: Intern
  • Fraunhofer Italia, Advanced Robotics Group, Bolzano, Italy
  • Supervisor: Dr. Andrea Giusti, Dr. Marco Todescato
  • Duties included: Building vision-guided imitation learning pipeline for modular robot manipulator, implementation of computer vision pipeline for 3D hand trajectory tracking, object recognition and segmentation with multi-stereo camera feed, re-implementation of Task-Parameterized Gaussian Mixture Model for imitation learning, development of Modular-Robot optimization pipeline, web-based interface design for containerized pipeline setup
  • Technologies used: Python, NumPy, OpenCV, ROS2, C++, Moveit, Eigen
• July 2022 - April 2023: Student Assistant
  • Institute of Intelligent Machine-Brain Interfacing Technology (IMBIT), Neurorobotics Group, University of Freiburg
  • Supervisor: Prof. Joschka Boedecker, Jasper Hoffmann
  • Duties included: Maintenance and mechanical work for Unitree A1 robot, interface documentation for custom quadrupedal policy deployment, literature review in reinforcement learning and control theory
• June 2021 - March 2023: Student Assistant
  • Fraunhofer Institute for Physical Measurement Technology (IPM), Object and Shape Detection Department
  • Supervisor: Prof. Alexander Reiterer, Dr. Philipp von Olshausen, Dr. Annette Schmitt
  • Duties included: Development of multilabel image classification pipeline with convolutional neural networks, conducted hyperparameter optimization study on fine-tuning of CNNs, collection of 3D point cloud data of urban habitation
• March 2020 - May 2021: Student Assistant
  • University Clinic Freiburg im Breisgau, Institute of Medical Bioinformatics and System Biology (IBSM)
  • Supervisor: Prof. Melanie Börries, Dr. Patrick Metzger, Maria Hess
  • Duties included: Developing a database application for molecular cancer research, conducting technical feasibility study involving relational and non-relational databases, implementing web interface for SQL database with Flask and SQLAlchemy, making application feasible for server-deployment in containerized fashion

Project Work

University Projects

• Master Thesis (since October 2025)
  • Lab of Prof. Georg Martius
  • Supervisors: Dr. Pavel Kolev, Núria Armengol Urpí, Marco Bagatella
  • Topics: occupancy modelling, generative models, offline meta-RL
• Distributed RL on Optimal Transport (April 2025 - October 2025)
  • Project at lab of Prof. Marcello Restelli, Politecnico di Milano
  • Supervisor: Dr. Marco Mussi
  • Redesign of rail environment from Flatland3 for multi-agent RL in asynchronous setting to solve train-scheduling problem
  • Switched perspective from train-focused to switch-focused
  • Tools: Python, Flatland3, Gymnasium, PettingZoo
• Field Robot Event (March 2025 - June 2025)
  • Multi-Disciplinary Project at Politecnico di Milano
  • Instructors: Prof. Matteo Matteucci, Dr. Simone Mentasti, Mirko Usuelli
  • Built and programmed robot for agricultural-robotics competition in team of nine
  • Third place at the Field Robot Event 2025 competition
  • Implementation of navigation, computer-vision and decision making algorithms
  • Conducted study on visual object tracking in moving stereo-camera setting with DeepSORT
  • Tools: Python, ROS2, PyTorch, Yolo
• Multi-Agent RL in Laser Hockey (December 2024 - February 2025)
  • Course Project on “Reinforcement Learning” taught by Prof. Georg Martius
  • Took part in an organized multi-agent RL competition in laser hockey
  • Extended SAC from Stable-Baselines3 with a custom replay buffer
  • Implemented and extended Deep Dyna-Q with entropy regularization
  • Contributed to development of the laser-hockey environment
• Generalized Simulation Based Inference (October 2024 - April 2025)
  • Practical Machine Learning, ML-Lab project
  • Supervisor: Prof. Jakob Macke, Julius Vetter
  • Extended sampling strategies in simulation-based inference with amortized cost estimation (paper by Dr. Richard Gao) with guided diffusion models
  • Re-implemented GBI-ACE for benchmarking on toy and real world datasets
  • Implementation using PyTorch framework
• LR-Scheduler Analysis (April 2024 - July 2024)
  • Seminar Project: “Deep Learning Research Kitchen”
  • Instructors: Prof. Philipp Hennig, Dr. Frank Schneider, Dr. Jonas Geiping, Dr. Antonio Orvieto
  • Conducted study of multiple learning rate schedulers and their hyperparameters affecting training of a Vision Transformer on CIFAR-10
  • Implemented pipeline using PyTorch-Lightning, TensorBoard, NumPy and Matplotlib
• Precipitation Prediction Analysis (December 2023 - February 2024)
  • Data Literacy Group Project by Prof. Philipp Hennig
  • Supervisor: Emilia Magnani
  • Data science project on climate data in group of four
  • Analyzed accuracy of precipitation forecast of German weather agency against real data for Baden-Württemberg
• EASRL-LM-2D-IKB (July 2023 - September 2023)
  • Bachelor Thesis at lab of Prof. Joschka Boedecker
  • Supervisor: Jasper Hoffmann
  • Topic: Expanding Action Space in Reinforcement Learning through Latent Models: A 2D Inverse Kinematics Benchmark
  • Merged Soft Actor-Critic with generative models such as pretrained Variational Autoencoder for noise augmentation and faster learning
  • Implementation of custom inverse kinematics environment with scalable action space
• Learning Quadrupedal Locomotion From Scratch (April 2022 - September 2022)
  • Bachelor research project with Felix Schulz at lab of Prof. Joschka Boedecker
  • Supervisor: Jasper Hoffmann
  • Implemented training pipeline for RL experiments around PPO concerned with perturbation and reward-shaping
  • Minimalist re-implementation of Vision-Guided Quadrupedal Locomotion
  • Used and extended existing physics-simulator based on PyBullet
  • Libraries: Stable-Baselines3, PyTorch, Gym, Weights & Biases, TensorBoard
  • Conducted study on closing sim-to-real gap on Unitree A1 robot
• ALE Plots (May 2022 - September 2022)
  • Seminar Project: “Fair and Interpretable Machine Learning”
  • Instructor: Dr. Janek Thomas
  • Re-implementation of Accumulated Local Effect plots
  • Implementation using Python with NumPy and Matplotlib
• Humboldt Game (October 2020 - April 2021)
  • Course project on “Practical Software Development”
  • Organizers: Prof. Andreas Podelski, Dr. Vincent Langenfeld
  • Supervisor: Antonia Hinkerohe
  • Designed, implemented and tested real-time multi-agent strategy game “Humboldt”
  • Collaborated in team of six using Git for version control
  • Programmed in C# using XNA-(monogame)-framework

Private Projects

• LazySlurm (since November 2025)
  • Terminal user interface for monitoring Slurm HPC jobs, published on PyPI as lazyslurm-py under the MIT license
  • Live overview of running and past jobs, log inspection, resource and CPU/GPU usage, job cancellation, editing and resubmission, and partition load
  • Remote mode running over a single background SSH connection with one-time two-factor authentication
  • Built in Python with the Textual framework
• Simplified RNNs Research Project (since April 2025)
  • Based on GateL0rd paper by Dr. Christian Gumbsch
  • Simplifying architecture and conducting study on latent state similarities
• SPARK Research Project (April 2025 - November 2025)
  • Stochastic Policies Augmented for Robust Knowledge/Exploration
  • Built on idea of Bachelor’s Thesis
  • Augmentation of stochastic policies for robust exploration and more expressive action distributions
  • Extension of SAC to leverage non-diagonal Gaussian distributions for action sampling
  • Benchmarking on 2D inverse kinematics environment using Stable-Baselines3
• Tide - Torch Integrated Data Engine (since February 2025)
  • Data toolkit for handling sequential data for single or multi-file datasets
  • Compatible with PyTorch dataset, data-loader integration, feature engineering, time-based splitting, filtering
  • Integration into publicly available time series data: EMG, climate, finance, robotics
• Config2Class (since October 2024)
  • Developed Python tool to automatically generate structured configuration classes from various file formats
  • Adds support for code completion in IDE
  • Integrates with Hydra and OmegaConf
• PyArgWriter (CLIfy) (since September 2023)
  • Developed Python tool to automatically create command-line interface for Python projects based on OOP paradigm
  • Implemented automatic code generation, code parsing tooling, tree-structured command generation, Hydra integration
• Hexapod Robot (since May 2021)
  • Building custom Hexapod robot using Freenove Big-Hexapod robot kit, Jetson Nano B01, stereo vision, custom stationary power supply
  • Implementing RViz2 and Gazebo simulation plugins, ROS2 architecture, low-level servo controller, high-level walking controller
  • Designing custom proprioceptive sensing for closed-loop controller design for cheap-PWM sensors
  • Conducted study on mobile neural-depth estimation for stereo-vision

Student Assistant Research Projects

• Human2Humanoid VLA Research Project (October 2025 - March 2026)
  • Lab of Prof. Georg Martius and Prof. Wieland Brendel
  • Supervisors: Anna Manasyan, Jens Beißwenger
  • Finetuning Vision-Language-Action models applied to humanoid robotics
  • Collecting humanoid demonstration data
  • Conducting studies on action representations and generalization in behavior cloning
  • Topics: tele-operation, re-targeting, RL, action representation, human-to-robot generalization
• Hand Kinematics Estimation from EMG Data (August 2024 - February 2025)
  • BG-Clinic project
  • Supervisor: Daniel Höglinger at lab of Dr. Cosima Prahm
  • Designed deep learning pipeline for hand kinematics estimation based on EMG data
  • Designed the pipeline for use by researchers outside computer science
  • Involved redesign of dataset acquisition and processing, network architecture setup, augmentation handling
  • Pipeline based on PyTorch-Lightning
  • Datasets: NinaPro, SEED, emg2pose
• PACT Deep Energy Retrofit (August 2022 - March 2023)
  • Fraunhofer IPM project under Dr. Annette Schmitt
  • Collected 3D data for point cloud segmentation with laser scanners
• Amphora Automated Pavement Operation (September 2021 - November 2022)
  • Fraunhofer-IPM project under Prof. Alexander Reiterer with Dr. Annette Schmitt
  • Implemented multilabel image classification pipeline on laser scan data with CNNs (ResNet & VGG) and Vision Transformer
  • Conducted classification-oriented real-time study with hyperparameter optimization (Bayesian Optimization + Hyperband)
  • Implemented image processing pipeline: laser scanner → geojson
• Molecular Tumor Board Research Database (March 2020 - May 2021)
  • IBSM team project under Prof. Melanie Börries
  • Development of database application for molecular tumor research
  • Conducted study on database design choices between SQL, document-based and graph-based database systems
  • Knowledge acquisition from tissue collection to sequencing to annotation pipeline
  • Collecting wet-lab experience and testing UI-design choices with research staff

Awards

• Field Robot Event - 3rd place (August 2025)
• System Design Project - 3rd place (February 2020)

Extracurricular Activities

• Private Statistics Tutor (September 2025 - October 2025)
  • Tutoring undergraduate students in statistics
• Organizational Staff - EWRL 18 (September 2025)
  • Volunteer at the 18th European Workshop on Reinforcement Learning in Tübingen
  • Assisting at registration, discussions, demonstrations
• Organizer - Machine Learning Reading Club (February 2025 - June 2025)
  • Organization of journal club for ML targeted at CS master students at Politecnico di Milano
  • Within association: “Polimi Data Scientists”
  • Tasks: selecting papers, organizing speakers, communicating to fellow students

Skills

• Programming Languages: Python, C, C++, C#
• Python Libraries:
  • Deep Learning: PyTorch, PyTorch Lightning
  • RL: Gymnasium, PettingZoo, Stable-Baselines3
  • Data Science: NumPy, Pandas, SciKit-Learn
• Robotics: ROS, ROS2, Moveit
• Databases: SQL, NoSQL Database Systems
• Tools: Git, SVN, Docker, SLURM
• CAD: Mechanical CAD (Autodesk Inventor)
• Embedded Systems: Arduino, ESP
• Project Management: SCRUM

Languages

• German (native)
• English (C1/C2)

Personal Interests

• Archery (bronze medal at European Championships 2018)
• Running (mid to long distance)
• Road cycling
• Photography