Flender

Practical tips for the coding, design, and behavior roundsPhoto by Pablo Heimplatz on UnsplashI recently landed an offer with Meta as Machine Learning Engineer (MLE), gaining a 20% raise in my total compensation relative to my previous job as well as a promotion to a senior role. In this post I’ll outline the interview structure, how I prepared, and practical tips on how to nail the MLE interview. Even though my experience was with Meta, I believe this guide to be useful for MLE interviews with other tech companies as…

Five lessons from the trenches of applied data sciencePhoto by Piret Ilver on UnsplashClass imbalance, where one class is much more abundant than the other, is one of the most ubiquitous topics in data science literature. Searching for ‘class imbalance’ on Medium alone reveals numerous articles with titles such as:“Dealing With Class Imbalanced Datasets For Classification”“How to Effortlessly Handle Class Imbalance with Python and SMOTE”“Stop Using SMOTE to Treat Class Imbalance”“A Loss Function Suitable for Class…

Forget ‘quiet quitting’: 3 strategies for creating more business impact with fewer hoursPhoto by Donald Wu on UnsplashDuring my career journey so far (first at JP Morgan Chase, then at Amazon), I’ve met some peers that appear to be constantly stressed out: they put in long hours and work hard, without creating that much real impact. On the flipside, I’ve also met peers that consistently deliver high impact to their organization without appearing stressed at all, all while working reasonable hours.How can this be? Why do…

Cookies and privacy, interleaving experiments, clean dial-ups, and test metricsPhoto by NASA on UnsplashA/B testing is one of the most critical steps in Machine Learning production: we only want to roll out a new ML model if it can be proven to be better in production. In Part I of this series we covered how to set up an A/B experiment with a population split, how to interpret the test results with statistical significance, and what kind of errors are to be expected. In this Part II we’ll go deeper into some practical…

How today’s tech companies make data-driven decisions in Machine Learning productionPhoto by Marcelo Leal on UnsplashA/B testing is is deeply ingrained in modern tech companies, enabling them to continuously improve their product in order to stay on top of consumer preferences and beat the competition. A Lyft article states that:The norm is to test each and every product change, to build up evidence to drive large decisions, and to use causal data to support the strategic direction.DoorDash writes:For any data-driven…

How to turn ML models into useful business actions: a primer on MLOpsPhoto by Katherine McCormack on UnsplashLike many people starting out in ML, one of the first problems that I got my hands on was the Titanic dataset. The task in this problem is to ‘predict’ whether a passenger survived the Titanic disaster or not, given features like ticket class, cabin location, gender, age, and so on. While it’s a fun problem to solve, it’s practically useless, for obvious reasons. No one actually needs a Titanic classifier. There’s…

Why ML models that look good on paper are not guaranteed to work well in productionPhoto by Florian Schmetz on UnsplashThese days, a typical ML research paper reads something like this:We propose a new model architecture X. We show that X outperforms SOTA by Y%. We conclude X is better than the current SOTA. Our code is available online.And that’s where academic research usually ends. However, from a production point of view, this is far from enough. There’s no guarantee that a model that looks good on paper actually…