Netherlands
Building analytical layers on top of high-throughput systems. Currently engaged FT with media company. Previously: Various services in the realm of automated algorithmic trading products and machine learning capabilities.
Apply machine learning to pricing and trading within various QR teams with trading desks. Also attached to Machine Learning Center of Excellence.
Responsible for setting up machine learning driven prediction products Built Tensorflow driven system to predict behavior in real time for all users active on Relay42’s client websites. System processed millions of raw user events daily into a custom Recurrent Neural Network that learns from past user behavior. Included building effective representations of complex, irregular and unstructured event data that preserve predict signal. Built automated model validation methods, result visualizations, hyperparameter search.
Lead of 3 person team responsible for technical implementation of a statistically driven home appraisal algorithm for the US real estate market. Algorithm uses data on more than 100 million homes to assess the current value of homes in the US.