Build a Career in Data Science (Emily Robinson, Jacqueline Nolis)（Z-Library）

Build a Career in Data Science EMILY ROBINSON AND JACQUELINE NOLIS M A N N I N G SHELTER ISLAND

For online information and ordering of this and other Manning books, please visit www.manning.com. The publisher offers discounts on this book when ordered in quantity. For more information, please contact Special Sales Department Manning Publications Co. 20 Baldwin Road PO Box 761 Shelter Island, NY 11964 Email: orders@manning.com ©2020 by Emily Robinson and Jacqueline Nolis. All rights reserved. No part of this publication may be reproduced, stored in a retrieval system, or transmitted, in any form or by means electronic, mechanical, photocopying, or otherwise, without prior written permission of the publisher. Many of the designations used by manufacturers and sellers to distinguish their products are claimed as trademarks. Where those designations appear in the book, and Manning Publications was aware of a trademark claim, the designations have been printed in initial caps or all caps. Recognizing the importance of preserving what has been written, it is Manning’s policy to have the books we publish printed on acid-free paper, and we exert our best efforts to that end. Recognizing also our responsibility to conserve the resources of our planet, Manning books are printed on paper that is at least 15 percent recycled and processed without the use of elemental chlorine. Development editor: Karen Miller Manning Publications Co. Review editor: Ivan Martinović 20 Baldwin Road Production editor: Lori Weidert PO Box 761 Copy editor: Kathy Simpson Shelter Island, NY 11964 Proofreader: Melody Dolab Typesetter: Dennis Dalinnik Cover designer: Leslie Haimes ISBN: 9781617296246 Printed in the United States of America

From Emily, to Michael, and From Jacqueline, to Heather, Amber, and Laura, for the love and support you provided us throughout this journey.

brief contents PART 1 GETTING STARTED WITH DATA SCIENCE .........................1 1 ■ What is data science? 3 2 ■ Data science companies 18 3 ■ Getting the skills 37 4 ■ Building a portfolio 55 PART 2 FINDING YOUR DATA SCIENCE JOB................................71 5 ■ The search: Identifying the right job for you 73 6 ■ The application: Résumés and cover letters 85 7 ■ The interview: What to expect and how to handle it 101 8 ■ The offer: Knowing what to accept 119 PART 3 SETTLING INTO DATA SCIENCE...................................135 9 ■ The first months on the job 137 10 ■ Making an effective analysis 155 11 ■ Deploying a model into production 174 12 ■ Working with stakeholders 192v

BRIEF CONTENTSviPART 4 GROWING IN YOUR DATA SCIENCE ROLE .....................213 13 ■ When your data science project fails 215 14 ■ Joining the data science community 228 15 ■ Leaving your job gracefully 246 16 ■ Moving up the ladder 261

contents preface xvii acknowledgments xix about this book xxi about the authors xxiv about the cover illustration xxvii PART 1 GETTING STARTED WITH DATA SCIENCE................1 1 What is data science? 3 1.1 What is data science? 5 Mathematics/statistics 6 ■ Databases/programming 7 Business understanding 9 1.2 Different types of data science jobs 10 Analytics 11 ■ Machine learning 11 ■ Decision science 12 Related jobs 13 1.3 Choosing your path 14 1.4 Interview with Robert Chang, data scientist at Airbnb 15 What was your first data science journey? 15 ■ What should people look for in a data science job? 16 ■ What skills do you need to be a data scientist? 16vii

CONTENTSviii2 Data science companies 18 2.1 MTC: Massive Tech Company 19 Your team: One of many in MTC 19 ■ The tech: Advanced, but siloed across the company 20 ■ The pros and cons of MTC 21 2.2 HandbagLOVE: The established retailer 22 Your team: A small group struggling to grow 22 ■ Your tech: A legacy stack that’s starting to change 23 ■ The pros and cons of HandbagLOVE 23 2.3 Seg-Metra: The early-stage startup 24 Your team (what team?) 25 ■ The tech: Cutting-edge technology that’s taped together 26 ■ Pros and cons of Seg-Metra 26 2.4 Videory: The late-stage, successful tech startup 28 The team: Specialized but with room to move around 28 The tech: Trying to avoid getting bogged down by legacy code 29 The pros and cons of Videory 29 2.5 Global Aerospace Dynamics: The giant government contractor 31 The team: A data scientist in a sea of engineers 31 ■ The tech: Old, hardened, and on security lockdown 32 ■ The pros and cons of GAD 32 2.6 Putting it all together 33 2.7 Interview with Randy Au, quantitative user experience researcher at Google 34 Are there big differences between large and small companies? 34 Are there differences based on the industry of the company? 35 What’s your final piece of advice for beginning data scientists? 35 3 Getting the skills 37 3.1 Earning a data science degree 38 Choosing the school 39 ■ Getting into an academic program 42 Summarizing academic degrees 43 3.2 Going through a bootcamp 44 What you learn 44 ■ Cost 46 ■ Choosing a program 46 Summarizing data science bootcamps 47 3.3 Getting data science work within your company 47 Summarizing learning on the job 49 3.4 Teaching yourself 49 Summarizing self-teaching 50

CONTENTS ix3.5 Making the choice 51 3.6 Interview with Julia Silge, data scientist and software engineer at RStudio 52 Before becoming a data scientist, you worked in academia; how have the skills learned there helped you as a data scientist? 52 ■ When deciding to become a data scientist, what did you use to pick up new skills? 53 ■ Did you know going into data science what kind of work you wanted to be doing? 53 ■ What would you recommend to people looking to get the skills to be a data scientist? 53 4 Building a portfolio 55 4.1 Creating a project 56 Finding the data and asking a question 56 ■ Choosing a direction 59 ■ Filling out a GitHub README 60 4.2 Starting a blog 60 Potential topics 60 ■ Logistics 61 4.3 Working on example projects 63 Data science freelancers 63 ■ Training a neural network on offensive license plates 64 4.4 Interview with David Robinson, data scientist 65 How did you start blogging? 66 ■ Are there any specific opportunities you have gotten from public work? 66 ■ Are there people you think would especially benefit from doing public work? 66 ■ How has your view on the value of public work changed over time? 66 ■ How do you come up with ideas for your data analysis posts? 67 ■ What’s your final piece of advice for aspiring and junior data scientists? 67 PART 2 FINDING YOUR DATA SCIENCE JOB......................71 5 The search: Identifying the right job for you 73 5.1 Finding jobs 74 Decoding descriptions 75 ■ Watching for red flags 77 Setting your expectations 77 ■ Attending meetups 78 Using social media 80 5.2 Deciding which jobs to apply for 81 5.3 Interview with Jesse Mostipak, developer advocate at Kaggle 83 What recommendations do you have for starting a job search? 83 How can you build your network? 83 ■ What do you do if you

CONTENTSxdon’t feel confident applying to data science jobs? 83 ■ What would you say to someone who thinks “I don’t meet the full list of any job’s required qualifications?” 84 ■ What’s your final piece of advice to aspiring data scientists? 84 6 The application: Résumés and cover letters 85 6.1 Résumé: The basics 86 Structure 88 ■ Deeper into the experience section: generating content 93 6.2 Cover letters: The basics 94 Structure 95 6.3 Tailoring 96 6.4 Referrals 97 6.5 Interview with Kristen Kehrer, data science instructor and course creator 99 How many times would you estimate you’ve edited your résumé? 99 What are common mistakes you see people make? 99 ■ Do you tailor your résumé to the position you’re applying to? 100 ■ What strategies do you recommend for describing jobs on a résumé? 100 What’s your final piece of advice for aspiring data scientists? 100 7 The interview: What to expect and how to handle it 101 7.1 What do companies want? 102 The interview process 103 7.2 Step 1: The initial phone screen interview 104 7.3 Step 2: The on-site interview 106 The technical interview 108 ■ The behavioral interview 111 7.4 Step 3: The case study 113 7.5 Step 4: The final interview 115 7.6 The offer 116 7.7 Interview with Ryan Williams, senior decision scientist at Starbucks 117 What are the things you need to do knock an interview out of the park? 117 ■ How do you handle the times where you don’t know the answer? 117 ■ What should you do if you get a negative response to your answer? 118 ■ What has running interviews taught you about being an interviewee? 118

CONTENTS xi8 The offer: Knowing what to accept 119 8.1 The process 120 8.2 Receiving the offer 120 8.3 Negotiation 122 What is negotiable? 122 ■ How much you can negotiate 125 8.4 Negotiation tactics 127 8.5 How to choose between two “good” job offers 128 8.6 Interview with Brooke Watson Madubuonwu, senior data scientist at the ACLU 129 What should you consider besides salary when you’re considering an offer? 130 ■ What are some ways you prepare to negotiate? 130 What do you do if you have one offer but are still waiting on another one? 130 ■ What’s your final piece of advice for aspiring and junior data scientists? 131 PART 3 SETTLING INTO DATA SCIENCE .........................135 9 The first months on the job 137 9.1 The first month 138 Onboarding at a large organization: A well-oiled machine 138 Onboarding at a small company: What onboarding? 139 Understanding and setting expectations 139 ■ Knowing your data 141 9.2 Becoming productive 144 Asking questions 145 ■ Building relationships 146 9.3 If you’re the first data scientist 148 9.4 When the job isn’t what was promised 149 The work is terrible 149 ■ The work environment is toxic 150 Deciding to leave 151 9.5 Interview with Jarvis Miller, data scientist at Spotify 152 What were some things that surprised you in your first data science job? 153 ■ What are some issues you faced? 153 ■ Can you tell us about one of your first projects? 153 ■ What would be your biggest piece of advice for the first few months? 154 10 Making an effective analysis 155 10.1 The request 158 10.2 The analysis plan 160

CONTENTSxii10.3 Doing the analysis 162 Importing and cleaning data 162 ■ Data exploration and modeling 164 ■ Important points for exploring and modeling 166 10.4 Wrapping it up 169 Final presentation 170 ■ Mothballing your work 171 10.5 Interview with Hilary Parker, data scientist at Stitch Fix 172 How does thinking about other people help your analysis? 172 How do you structure your analyses? 172 ■ What kind of polish do you do in the final version? 172 ■ How do you handle people asking for adjustments to an analysis? 173 11 Deploying a model into production 174 11.1 What is deploying to production, anyway? 175 11.2 Making the production system 177 Collecting data 178 ■ Building the model 178 Serving models with APIs 179 ■ Building an API 180 Documentation 182 ■ Testing 183 ■ Deploying an API 184 Load testing 187 11.3 Keeping the system running 187 Monitoring the system 187 ■ Retraining the model 188 Making changes 189 11.4 Wrapping up 189 11.5 Interview with Heather Nolis, machine learning engineer at T-Mobile 189 What does “machine learning engineer” mean on your team? 189 What was it like to deploy your first piece of code? 190 ■ If you have things go wrong in production, what happens? 190 ■ What’s your final piece of advice for data scientists working with engineers? 191 12 Working with stakeholders 192 12.1 Types of stakeholders 193 Business stakeholders 193 ■ Engineering stakeholders 194 Corporate leadership 195 ■ Your manager 196 12.2 Working with stakeholders 197 Understanding the stakeholder’s goals 197 ■ Communicating constantly 199 ■ Being consistent 201

CONTENTS xiii12.3 Prioritizing work 203 Both innovative and impactful work 204 ■ Not innovative but still impactful work 205 ■ Innovative but not impactful work 205 ■ Neither innovative nor impactful work 206 12.4 Concluding remarks 206 12.5 Interview with Sade Snowden-Akintunde, data scientist at Etsy 207 Why is managing stakeholders important? 207 ■ How did you learn to manage stakeholders? 207 ■ Was there a time where you had difficulty with a stakeholder? 207 ■ What do junior data scientists frequently get wrong? 208 ■ Do you always try to explain the technical part of the data science? 208 ■ What’s your final piece of advice for junior or aspiring data scientists? 208 PART 4 GROWING IN YOUR DATA SCIENCE ROLE...........213 13 When your data science project fails 215 13.1 Why data science projects fail 216 The data isn’t what you wanted 217 ■ The data doesn’t have a signal 218 ■ The customer didn’t end up wanting it 220 13.2 Managing risk 221 13.3 What you can do when your projects fail 222 What to do with the project 223 ■ Handling negative emotions 224 13.4 Interview with Michelle Keim, head of data science and machine learning at Pluralsight 226 When was a time you experienced a failure in your career? 226 Are there red flags you can see before a project starts? 226 How does the way a failure is handled differ between companies? 226 ■ How can you tell if a project you’re on is failing? 227 ■ How can you get over a fear of failing? 227 14 Joining the data science community 228 14.1 Growing your portfolio 230 More blog posts 230 ■ More projects 231 14.2 Attending conferences 231 Dealing with social anxiety 234 14.3 Giving talks 235 Getting an opportunity 236 ■ Preparing 239

CONTENTSxiv14.4 Contributing to open source 239 Contributing to other people’s work 240 ■ Making your own package or library 241 14.5 Recognizing and avoiding burnout 242 14.6 Interview with Renee Teate, director of data science at HelioCampus 243 What are the main benefits of being on social media? 243 ■ What would you say to people who say they don’t have the time to engage with the community? 244 ■ Is there value in producing only a small amount of content? 244 ■ Were you worried the first time you published a blog post or gave a talk? 244 15 Leaving your job gracefully 246 15.1 Deciding to leave 247 Take stock of your learning progress 247 ■ Check your alignment with your manager 248 15.2 How the job search differs after your first job 250 Deciding what you want 250 ■ Interviewing 251 15.3 Finding a new job while employed 252 15.4 Giving notice 254 Considering a counteroffer 255 ■ Telling your team 255 Making the transition easier 257 15.5 Interview with Amanda Casari, engineering manager at Google 258 How do you know it’s time to start looking for a new job? 258 Have you ever started a job search and decided to stay instead? 258 Do you see people staying in the same job for too long? 258 Can you change jobs too quickly? 259 ■ What’s your final piece of advice for aspiring and new data scientists? 259 16 Moving up the ladder 261 16.1 The management track 263 Benefits of being a manager 264 ■ Drawbacks of being a manager 264 ■ How to become a manager 265 16.2 Principal data scientist track 267 Benefits of being a principal data scientist 268 ■ Drawbacks of being a principal data scientist 269 ■ How to become a principal data scientist 270

CONTENTS xv16.3 Switching to independent consulting 271 Benefits of independent consulting 272 ■ Drawbacks of independent consulting 272 ■ How to become an independent consultant 273 16.4 Choosing your path 274 16.5 Interview with Angela Bassa, head of data science, data engineering, and machine learning at iRobot 275 What’s the day-to-day life as a manager like? 275 ■ What are the signs you should move on from being an independent contributor? 275 ■ Do you have to eventually transition out of being an independent contributor? 275 ■ What advice do you have for someone who wants to be a technical lead but isn’t quite ready for it? 276 ■ What’s your final piece of advice to aspiring and junior data scientist? 276 Epilogue 280 appendix Interview questions 282 index 311

preface “How do I get your job?” As veteran data scientists, we’re constantly being asked this question. Sometimes, we’re asked directly; at other times, people ask indirectly through questions about the decisions we’ve made in our careers to get where we are. Under the surface, the peo- ple asking the questions seem to have a constant struggle, because so few resources are available for finding out how to become or grow as a data scientist. Lots of data sci- entists are looking for help with their careers and often not finding clear answers. Although we’ve written blog posts with tactical advice on how to handle specific moments in a data science job, we’ve struggled with the lack of a definitive text cover- ing the end-to-end of starting and growing a data science career. This book was written to help these people—the thousands of people who hear about data science and machine learning but don’t know where to start, as well as those who are already in the field and want to understand how to move up. We were happy to get this chance to collaborate in creating this book. We both felt that our respective backgrounds and viewpoints complemented each other and cre- ated a better book for you. We are ■ Jacqueline Nolis—I received a BS and MS in mathematics and a PhD in opera- tions research. When I started working, the term data science didn’t yet exist, and I had to figure out my career path at the same time that the field was defining itself. Now I’m a consultant, helping companies grow data science teams. ■ Emily Robinson—I got my undergraduate degree in decision sciences and my master’s in management. After attending a three-month data science bootcampxvii

PREFACExviiiin 2016, I started working in data science, specializing in A/B testing. Now I work as a senior data scientist at Warby Parker, tackling some of the company’s biggest projects. Throughout our careers, we’ve both built project portfolios and experienced the stress of adjusting to a new job. We’ve felt the sting of being rejected for jobs we wanted and the triumph of seeing our analyses positively affect the business. We’ve faced issues with a difficult business partner and benefited from a supportive mentor. Although these experiences taught us so much in our careers, to us the true value comes from sharing them with others. This book is meant to be a guide to career questions in data science, following the path that a person will take in the career. We start with the beginning of the journey: how to get basic data science skills and understand what jobs are actually like. Then we go through getting a job and how to get settled in. We cover how to grow in the role and eventually how to transition up to management—or out to a new company. Our intention is for this book to be a resource that data scientists continue to go back to as they hit new milestones in their careers. Because the focus on career is very important for this book, we chose to not focus deeply on the technical components of data science; we don’t cover topics such as how to choose the hyperparameters of a model or the minute details of Python pack- ages. In fact, this book doesn’t include a single equation or line of code. We know that plenty of great books out there cover these topics; we wanted instead to discuss the often-overlooked but equally important nontechnical knowledge needed to succeed in data science. We included many personal experiences from respected data scientists in this book. At the end of each chapter, you’ll find an interview describing how a real, human data scientist personally handled dealing with the concepts that the chapter covers. We’re extremely happy with the amazing, detailed, and vulnerable responses we got from all the data scientists we talked to. We feel that the examples they provide from their lives can teach much more than any broad statement we might write. Another decision we made in writing this book was to make it opinionated. By that, we mean we intentionally chose to focus on the lessons we’ve learned as professional data scientists and by talking to others in the community. At times, we make state- ments not everyone might agree with, such as suggesting that you should always write a cover letter when applying for jobs. We felt that the benefit of providing viewpoints that we strongly believe are helpful to data scientists was more important than trying to write something that contained only objective truths. We hope that you find this book to be a helpful guide as you progress in your data science career. We’ve written it to be the document we wish we had when we were aspiring and junior data scientists; we hope that you’ll be glad to have it now.

acknowledgments First and foremost, we’d like to thank our spouses, Michael Berkowitz and Heather Nolis. Without them, this book would not have been possible (and not just because Michael wrote the first draft of some of the sections despite being a bridge profes- sional and not a data scientist, or because Heather evangelized half of the machine learning engineering content). Next, we want to acknowledge the staff at Manning who guided us through this process, improved the book, and made it possible in the first place. Thank you espe- cially to our editor, Karen Miller, who kept us on track and coordinated all the vari- ous moving parts. Thank you to all the reviewers who read the manuscript at various points and pro- vided invaluable detailed feedback: Brynjar Smári Bjarnason, Christian Thoudahl, Daniel Berecz, Domenico Nappo, Geoff Barto, Gustavo Gomes, Hagai Luger, James Ritter, Jeff Neumann, Jonathan Twaddell, Krzysztof Jędrzejewski, Malgorzata Rodacka, Mario Giesel, Narayana Lalitanand Surampudi, Ping Zhao, Riccardo Marotti, Richard Tobias, Sebastian Palma Mardones, Steve Sussman, Tony M. Dubitsky, and Yul Wil- liams. Thank you as well to our friends and family members who read the book and offered their own suggestions: Elin Farnell, Amanda Liston, Christian Roy, Jonathan Goodman, and Eric Robinson. Your contributions helped shape this book and made it as helpful to our readers as possible. Finally, we want to thank all of our end-of-chapter interviewees: Robert Chang, Randy Au, Julia Silge, David Robinson, Jesse Mostipak, Kristen Kehrer, Ryan Williams, Brooke Watson Madubuonwu, Jarvis Miller, Hilary Parker, Heather Nolis, Sade Snowden-xix