Think Before You Build and Fake It Till You Make It

aka. Lessons Learned While Building Machine Learning Product from Scratch.

My favourite part of being a product manager is definitely product discovery. I’m passionate about talking to customers, uncovering their problems and needs, and then trying to solve them with the right product. Since I joined Zalando in 2016, I’ve been focussing on machine learning-based products and trying to marry discovery-driven product development with the complexities of machine learning research lifecycle. In my talk, I’d like to walk the audience through a case study of how I built an entirely new product at Zalando: outfit recommendations, which help our customers to match their purchases with new items and create personalized outfits, and what I’ve learned during that journey: starting from the importance of solving the right problem, to various ways of how one can cheaply “fake” machine learning solutions, to what I understand by embracing a “machine learning mindset” by a PM.