About Me

Hey there,

I am a M.Sc. process engineer who specializes in the field of Machine Learning (ML). For this purpose I studied computer science at the University of Hamburg for one year which gave me enough time to learn Python and how to work with data. I now consider myself as a junior data scientist with some first practical experience in different fields and applications (see below for details). 
I’ve proven my skills with a high grade in the master course “Machine Learning” (at Uni Hamburg) and by winning 1,5k Euro at the SeaDevCon Hackathon in the “Zero Emission” Challenge sponsored and supported by Wärtsilä. 

If you want to work with me:

I have experience with a lot of Machine Learning techniques, including classical approaches like SVM’s and Random Forrest, as well as currently popular algorithms including Deep Learning, Convolutional- and Recurrent Neuronal Networks (NN’s). I have worked in the field of Computer Vision (CV), Predictive Maintenance, Classification and Clustering. As a good data scientist I’m of cause able to clean the data beforehand and find the right model for the specific purpose that is given to me.

Contact me via LinkedIn or send me an Email to job@predicted.blog

An excerpt of my philosophies in ML:

  • Don’t do NN’s first: I’m convinced that most of the time (except for the field of computer vision) Neuronal Networks are not the best model to begin with. The reason for that consists of two parts: 1. They have a lot of hyper parameters like network architecture, learning rate etc. that need to be tuned, so it takes a lot of time to get them to work properly. 2. Neuronal Nets are Black Box models, so most of the time you don’t exactly know why it’s producing certain results. If there is enough time and data however, they most certainly “outperform” the classical approaches.
  • Garbage in Garbage out: Of cause the Model plays an important role in achieving awesome results, but even more important is high quality training data. Every algorithm can only perform as good as the data it’s given. In my experience companies usually have incomplete and messy data so it’s important to perform an effortful cleaning and feature engineering process before you even start modeling. This usually takes about 70-80 % of the time in a project.
  • Learn from others: To work in the field of ML it’s equally important to stay up to date with modern algorithms as well as looking explicitly for solutions to similar problems as the one you are currently facing. It’s not enough to rely on what you have learned at university or in an online ML course because the field is rapidly changing every year.