I am driven by the goal to turn the fascinating potential of artificial intelligence into valuable solutions for customers, companies and society. I have a twofold academic background in management and computer science. In my computer science degree at TUM I specialised on artificial intelligence, machine learning and natural language processing. In my management degree at LMU Munich I focused on organisational design, leadership and value-based management. In order to fully exploit the potential of modern technologies it is crucial to develop a solution that fits both worlds. I gained experience in rapid prototyping and building MVPs in several projects in the areas NLP, Healthcare, Computer-Aided Diagnosis and Explainable AI.
B.Sc. Business Administration | Value Based Management | Evaluation | Strategic Organization Design
M.Sc. Computer Science | B.Sc. Computer Science | Artificial Intelligence | Natural Language Processing | Data Science | Explainable AI
Sailing - Skipper at Sailwithus | (Social) Entrepreneurship - Enactus | Traveling | Food | Cooking | Drone
Technology
Many administrative tasks in banks and incusrances are still based on paper-based forms. Digitalizing this data is a tedious task. I implemented an OCR prototype with Dataiku to automatically process those forms and extract the information automatically. The model also incorporates features for handwritten text recognition.
Technology
Explainability of artificial neural networks is crucial to fully unlock their potential in many application. Siamese Neural Network have special challenges when it comes to explainability. The research in my master thesis is the first existing work showing how common explainability techniques can be applied on this architecture. The solution is shown in the case of computer aided diagnosis. I developed a solution that automatically detects pneumonia in x-ray images and provides an explanation to the doctors of where the inflammation is assumed by the model.
Technology
Artificial neural networks are powerful in many medical applications. However, in a lot of cases sufficiently large datasets are not available. I developed a GAN to generate artificial training data. It significantly improves the performance of supervised downstream tasks like classification or regression problems. The GAN was trained on the medMNIST Dataset of x-ray chest scans.
Technology
Automatically identifying, understanding and summarizing the core topics in online debates is an unpayable value for companies and political parties. The public opinion to products, solutions and political campaigns is at the core of their success. I developed a solution to automatically summarize arguments in social media with corresponing key points. Several developed explainability techniques give the user the ability to understand the rationals between a trained siamese neural network.
Technology
Autonomous Agents are the future. Robots find their way into our daily live. No matter whether they deliver post, drive taxi or perform caring tasks in hospitals.. Those autonomous agents need to succesfully move in our environment. Therefore they need collision avoidance mechanisms. However, this is not a trivial problem as we have to deal with a spatio-temporal space. I developed a new Algorithm called PCMP which allows to control the risk aversion of an Agent. It is a trade-off between the fastest route to the destination or the risk of collisions. The PCMP allows to control for this risk aversion and can be used as a high level collision management procedure. The algorithm was published in the SIGSPATIAL 21.
Technology
NLP-Web-Application to transform the way we work together. Together with the research Chair of HR Management at the LMU Munich, I developed a solution to solve the "who knows what problem" in large organizations. The solution automatically creates knowledge profiles for employees based on their daily work. It connects allows knowledge to flow in the organization overcoming hierarical, geographic and personal borders.
German
100%
English
80%
Spanish
85%
Python
80%
Java
50%
JavaScript
30%
SQL
30%
Tailwind
30%
Tensorflow
40%
Computer Vision
40%
React
20%
PyTorch
20%
Algorithms
60%
Keras
60%
Natural Language Processing
50%
TypeScript
30%
GitHub
50%
Office
80%
Next
20%
Artificial Intelligence
60%
Dataiku
40%
Python
70%
+491747771782
hannes@schroter.biz
Heidelberg
