About me: Konstantin Ditschuneit #

LinkedIn | Github | ko.ditschuneit@gmail.com | CV

Welcome! I’m a passionate Machine Learning Researcher and Engineer. Below you’ll find out about my work experience, projects I enjoyed working on and my education.

Work Experience #

Founding deep learning engineer at Scenarium AI #

2023 - Present

In my current role, I:

• Develop tooling for electrical planners to automate repetitive parts of their daily work,
• Build a dataset of 3d building plans to train Machine Learning models despite a lack of publicly available data or pretrained models,
• Joined as the first employee to build a product from scratch.

Machine learning researcher at Merantix Momentum GmbH #

2021 - 2023

At Merantix Momentum, I

• Worked on a number of publicly funded research project conducting research on semantic segmentation, interpretable machine learning, and self-supervised time-series learning.
• Lead and managed the design, maintenance, and enhancement of the research cloud infrastructure for 15 researchers.

Mobile application developer at Poolarserver GmbH #

2016

Right after school, I joined Poolarserver for an internship and developed the mobile app PoolarOffice from ground up.

Notable Projects #

Rustgrad #

2025

Together with a friend, I created rustgrad, a little machine learning package written in rust. It features type and shape safety, automatic differentiation, and stochastic gradient descent. Check it out on Github at

Pytorch-tsmixer #

2023

As a fun project, I created pytorch-tsmixer, an implementation of the TSMixer: An all-MLP architecture for time series forecasting publication in pytorch.

KITcar e.V. #

2019 - 2021

I was fortunate to lead the “simulation” team at KITcar e.V., guiding a team of 5+ students to develop a 3D simulator for vehicle testing. Highlights from this project include:

• Founding and leading a team of 5+ students focused on simulation.
• Developing and open-sourcing a ROS-based 3D simulator capable of fully testing our vehicles in a virtual environment.
• Implementing a Generative Adversarial Network (GAN) that transformed simulated images into photo-realistic ones.
• Winning the prestigious Carolo-Cup competition in both 2020 and 2021.

Education #

Karlsruhe Institute of Technology #

Master of Science in Computer Science, 2019 - 2022 | Final Grade: 1.5

• Deep Learning, Algorithm Engineering and Efficient Computing
• Thesis: Increasing robustness by using Mixture-of-Expert models

Humboldt-Universität zu Berlin #

Bachelor of Science in Physics, 2016 - 2019 | Final Grade: 1.6

• Tutor for Linear Algebra
• Thesis: Searching for algebraic structures in Multiple-Elliptic-Polylogarithms