How to best prepare for a DS Interview at startups of any round of funding, internships or full time

I’ve spent my career as a Data Scientist at startups. I’ve also been a DS hiring manager for small teams along the way.

People make a lot of mistakes in startup interviews.

Photo by Steve Halama on Unsplash

Today, there is still a huge hiring boom in Data Science — startups are desperate for qualified Data Scientists & Machine Learning Engineers.

However, I’ve seen an alarming number of DS candidates highlighting the wrong things on resumes and in interviews, which has led us to choose another candidate.

I wanted to briefly touch on each of the common traps DS candidates fall into, and ways you can…

A Quick Discussion on Minds & Metaphors… and running electricity through our amygdala

Lions: The King of the Jungle.

Blood Vessels: The Highways of the human body.

Mitochondria: The powerhouse of the cell.

Modern science is inundated with narrative and metaphor. And with good reason: it is impossible for us mere humans to experience much of the world directly. Metaphors play a vital role in our capacity to quickly relate to these foreign concepts.

For the most part, metaphors are a fantastic tool. …

How Hedging your Fears can hopefully buy you some piece of mind

**Disclaimer: Politics is inherently a controversial topic, and Politics + Math don’t usually mix well. Please realize this is a casual article, not a treatise on political science…just thought it was fun idea to talk about!**

You don’t need to be following American Politics closely to know that last November’s U.S Presidential Election was full of strong emotions. Both candidates garnered such loathing that opposition on both sides descried them as a the beginning of the end of America as we know it.

As usual, it is a hotly contested battle for the Oval Office. Photo by Srikanta H. U on Unsplash

Given that both sides fear the consequences of their candidate losing, it is important to ask the…

How to use KMeans & determine how many clusters to use in your analysis

Clustering is a fundamental skill in your Data Science toolkit. It can solve a huge array of problems — from user segmentation to anomaly detection — and can help your team derive very interesting insights.

Photo by Chris Ried on Unsplash

Determining the right number of clusters for your project is a little more art than science. In this article, I will go over a few common ways to determine the right number of clusters.

All of the following examples will involve the following imports:

import pandas as pd
import numpy as np
import matplotlib.pyplot as plt
from sklearn.cluster import KMeans

and the following variables:


All Shakes and No Snacks make Adam a Dull Boy

Photo by Joseph Greve on Unsplash

Spoiler: It was not fun.

But there were some perks, and definitely some great learnings on consistency and discipline. Hope you enjoy!


Sitting around a (virtual) dinner table, my friends and I were discussing how COVID had impacted both our social lives and our eating habits.

Heather: “It’s strange how big a part socializing plays on your diet and the foods you eat… delivery food doesn’t taste as good as the same food when I used to eat it out with friends.”

Me: “I wonder if I didn’t socialize at all, I could eat the same meal over and over…

How to get an offer to join Facebook’s Data Science Team

This past recruiting cycle, I was lucky enough to participate in Facebook’s DS Interview process from initial phone screen to offer. Here is my experience and some tips I’d give anyone else going through this rather lengthly experience in the near future (or right now!)

Photo by on Unsplash

Phone Screen #1: Behavioral Screen

Like most Data Science/Analyst positions, the initial call is with a recruiter to asses my relevant background and skills. Pretty typical questions like:

  • Tell me about a recent project you worked on involving Machine Learning
  • How has your experience at [company] prepared you for this role?

Phone Screen #2: Product Sense Screen

Unlike most Data Science/Analyst positions, the second phone call…

A how-to guide for Design Sprinting Products as fast as humanely possible

Photo by Anukrati Omar on Unsplash

Are you building products without knowing if anyone will use it? Are you finding it takes forever to do market validation? Are you part of a bootstrapped/scrappy start-up? If so, then this post is for you.

Speed Running a Design Sprint

A Tech Sprint is a framework for rapidly and efficiently testing new ideas to see if they would make great products. At my start-up when we first needed a product validation framework, we had a long list of exciting potential ideas. We also had very limited time and resources to do so. So we made it our goal to:

“Effectively validate our product ideas…

My Data Science Love Letter to 7/8ths of Game of Thrones

Unfortunately, last night I watched Game of Thrones season 8 episode 3.

Author’s Note: While the source material has fallen out of contemporary news, my fiery passion for Game of Thrones, and my anger surrounding the abysmal collapse of the show in season 8 have not subsided. Here is my love letter to a show I loved dearly for 7 years and hated ever after.

** Citations marked with superscript numbers **


On the spectrum of Game of Thrones fandom, I imagine I fall somewhere in the upper-middle. …

Not All Customer Models Are Created Equal

We can all agree that COVID has completely changed customer behavior. More people are buying digitally, more people are stuck at home, and more people are investing in personal comforts over trips/vacations.

And if they are anything like me, more people are going on weekend-long Amazon shopping sprees…

The FedEx mailman is my best friend at this point.

Photo by Elena Mozhvilo on Unsplash

Even with all of these changes (and particularly because of these changes) we need to ensure that our business can understand our customer base and continue to grow in these turbulent times.

Developing a powerful and accurate long-term value model for your…

A brief intro to the BTYD family, Pareto/NBD, & Pareto/GGG for Predicting Buying Behavior

Predicting a Customer’s Lifetime Value, or how much they will spend over the next few years, is a very challenging problem. Some Data Scientists (me) have full time jobs dedicated to solving this problem.

Photo by freestocks on Unsplash

Luckily, there is a family of models specifically built to address this tricky problem: BTYD Models!

In this article, I hope you’ll learn (1) what are BTYD Models and (2) conceptually how to the most common BTYD Models work under the hood.

Let’s get started!

Note: In my naive opinion, the BTYDplus R Package is the de-facto best library in any language to explore BTYD models…

Adam Brownell

Sr. Data Scientist & AI PM || Writing about ML/AI, Product, and the Cognitive Sciences. Love telling Data Stories!

Get the Medium app

A button that says 'Download on the App Store', and if clicked it will lead you to the iOS App store
A button that says 'Get it on, Google Play', and if clicked it will lead you to the Google Play store