This is my personal blog where I’ll be posting some random stuff about machine learning and data science.
I’ve been obsessing about Time To Event modeling since the days of Theano. I just moved to Silicon Valley after quitting my job as a Staff Machine Learning Engineering at Schibsted where I worked with the problem of predicting things for millions of users every day and all the lovely engineering challenges that comes with making ML products and teams work well.
I studied Engineering Mathematics at Chalmers University of Technology where I focused on ML and optimization.
Feel free to gmail me about anything egil.martinsson@ or linkedin.
I think the best way to explain what ML-algos does and the kind of output they produce is to start with what it was taught to learn in the first place. This is solely defined by the loss/objective function we optimize over. What’s interesting here is the kind of trade-offs that the objective function induces. I like the perspective of seeing machine learning as mathematical optimization sparkled with probablistic stories. I also think it’s a good modeling-habit. Hence I’ll try to focus on the objective.
Site is built using lanyon theme for Jekyll and uses google analytics