Welcome to My Portfolio
I'm a data scientist and machine learning enthusiast with a passion for learning new technologies and building projects.
Skills
These are my most useful technological skills and the languages i speak.
Technologies
Python
Pandas
Polars
Flask
Linux / Bash
Git & CI/CD
PostreSQL
Docker
MLflow
NumPy
Plotly
Scikit-learn
Google Cloud
Docker
PyTorch
Prometheus + Grafana
Java
Terraform
HTML, JavaScript, CSS
Kubernetes
Languages
Projects
-
2020 - today Running and Maintaining a private Nextcloud on a RaspberryPi
Installation and regular maintenance of a Nextcloud instance hosted locally on a RaspberryPi that is accessible through the public internet. Developed a custom backup solution that regularly syncs the data to a second micro computer over a VPN tunnel in a different location.
-
September 2024 ESP32 E-Ink for Displaying Weather and Public Transportation Information
Paired an e-ink display with an ESP32 that shows weather and public transportation informations in an image frame. A locally hosted FastAPI server aggregates the relevant information from public APIs.
-
2020 - today Running and Maintaining a private Nextcloud on a RaspberryPi
Installation and regular maintenance of a Nextcloud instance hosted locally on a RaspberryPi that is accessible through the public internet. Developed a custom backup solution that regularly syncs the data to a second micro computer over a VPN tunnel in a different location.
-
September 2024 ESP32 E-Ink for Displaying Weather and Public Transportation Information
Paired an e-ink display with an ESP32 that shows weather and public transportation informations in an image frame. A locally hosted FastAPI server aggregates the relevant information from public APIs.
-
November 2023 Python Programming Exercise for First-Semester Students
Developed a comprehensive Python programming exercise for first-semester students where they had to implement simple encryption and decryption algorithms.
-
March 2023 User Authentication for Kubeflow with AD over Keycloak
Integrated Active Directory as an authentication method for Kubeflow using Keycloak.
-
November 2022 VRcher
Created a VR archery game in Unity using C#.
-
June 2021 ImageCoder
Built a steganography tool that encodes a message in an image with Java.
-
November 2020 ESP8266 Air Sensor
Designed and built an ESP8266-based temperature, humidity and air pressure logger using C++. Built a custom back- and frontend using PHP.
Work Experience
Working Student Data Management and Analysis
inovex GmbH, September 2022 - Present
- Working at the research project DeKIOPs for the democratization of AI Operations.
- Creating prediction models to reduce battery waste in sensor logging devices.
- Using Pandas, Polars, Scikit-learn, Scikit-survival, Flask, Plotly.
Tutor for Python Programming
Hochschule Karlsruhe - University of Applied Sciences, October 2023 - February 2024
- Helped data science students in their first semester to learn Programming in Python.
Practical Semester Data Management and Analysis
inovex GmbH, April 2022 - August 2022
- Worked in the research project "Service-Meister", helping to create a predictive
maintenance platform.
- Used Pandas, Scikit-learn and NLTK to categorize and cluster the maintenance descriptions.
- Used Flask, MlFlow, Docker, Terraform, Gitlab CI/CD, PostgreSQL to build a backend
infrastructure.
Tutor for Java Programming
Hochschule Karlsruhe - University of Applied Sciences, October 2019 - February 2022
- Helped computer science students in their first semester to learn Programming in Java.
Education
Master's Degree in Computer Science focused on Machine Learning
Hochschule Karlsruhe - University of Applied Sciences, October 2024 - present
Bachelor's Degree in Computer Science
Hochschule Karlsruhe - University of Applied Sciences, March 2019 - June 2024
Thesis: "Optimization of Prediction Models for Battery Life Forecasting in IoT Devices using Augmented Data"
Final grade: 1.4
A-Levels
Gymnasium in der Taus Backnang, September 2010 - June 2018
Final grade: 2.0
Primary School
Mörikeschule Backnang, August 2006 - August 2010
Publications & Conferences
-
2025 Predicting Battery Discharge Behavior for Sparse Data with Augmentation
Co-authored with colleagues at inovex GmbH. Presented at IFIP International Conference on Artificial Intelligence Applications and Innovations. We developed a comprehensive workflow for predicting battery lifetimes in IoT devices using sparse multivariate time series. The methodology combined data acquisition, augmentation, preprocessing, feature engineering, and modeling. By applying augmentation techniques, we improved the Integrated Brier Score by up to 50.16% and introduced a novel evaluation metric, Mean Divergence Time, showing predictions remain reliable for an average of 117.76 days. This work demonstrates how augmented data can significantly reduce costs and ecological impact in battery-dependent systems.
Read Paper