Senior Data Scientist2018 - PRESENTAva AG
Technologies: Amazon Web Services (AWS), TensorFlow, Scikit-learn, Python
- Led the data science contribution to a novel product for detecting pregnancy anomalies, additionally resulting in a new patent application.
- Developed a sleep stage classification algorithm based on physiological data.
- Performed user data mining for new insights and scientific publications.
Co-founder | CTO2016 - PRESENTKlapp
- Acted as the co-founder and CTO of an edtech startup developing a parent-teacher communication platform.
Data Scientist2016 - 2018Swiss Re
Technologies: Spark, R, Python
- Developed the weather model for a statistical/ML based flight delay insurance pricing engine.
- Built a probabilistic global tsunami risk model.
- Created an ML-based predictive model for insurance premiums and claims.
Quantitative Analyst2013 - 2016MET International AG
- Created a web-based commodity trading interface on top of a forward price curve generation algorithm for the straight-through order placement and processing of natural gas deals using Python for the numerical computation, STP, and web back-end and HTML, Bootstrap, and AngularJS for the front-end.
- Developed a Monte Carlo-based method for the optimal pricing and hedging of natural gas storage and implemented it in Python (NumPy, Pandas).
- Researched and developed a spread-based trading strategy on the German power markets and traded it live on the EEX.
- Developed a detailed physical model for a natural-gas-fired power plant and used it to derive optimal hedging and trading strategies.
Quantitative Analyst2010 - 2013swissQuant Group AG
Technologies: Java, MATLAB
- Developed a novel (non-heuristic) method for achieving higher asset diversification within the context of optimal portfolio construction and implemented it using MATLAB (computation) and Java (front-end).
- Wrote a method for optimal hedging of currency exposure using arbitrary linear and non-linear instruments (e.g., options), based on normal or non-normal underlying distribution assumptions, and implemented a prototype using MATLAB and Java.
- Created a method for tracking a non-investable insurance-linked securities index using a set of arbitrary investable products by modeling the index as a jump-diffusion process and minimizing the CVaR of the performance difference.
- Developed a method for hedging the future power generation of a wind farm connected to a liquid power market using liquid instruments.