Lukas Ehrig

Developed a Machine learning-based screening application to calculate the risk of fetal alcohol spectrum disorder among ADHD patients in conjunction with my master’s thesis. Will be used in practice and evaluated in a prospective study. Publication pending.

Created a method for predicting the Remaining Useful Life (RUL) of turbofan engines using time series analysis on NASA aircraft sensor data, incorporating the real business costs for under- and overestimation of the RUL . Best paper award, published at IEEE.

Improved and extended capabilities of conversational AI platform at the SAP Leonardo conversational AI team. Responsible for triggering and executing the correct dialog flow after intent matching and using relevant information extracted by named entity recognition. I also helped in preparing language data for machine learning and exploring new AI technologies. During the feature freeze, I was involved in the backend and frontend testing and fixing.

Applied machine learning techniques on millions of historical sales orders to identifiy the characteristics of sales orders whose fulfllment was problematic. As part of my bachelor’s project team, I co-created a predictive analytics platform that utilizes this knowledge to calculate the probability of issues to occur for a new sales order, giving the reasons for its assessment and suggesting the most effective action to mitigate the issues.