✂️ Breaking into Data Science — Applying & Interviewing (Ch.3)- 🔍
Tl:dr 💁️ ➡️ 👩💻;
How I went from hair salon girl to data scientist working for an innovative digital health company with no Master’s, PhD, or quantitative college degree. (Chapter 3)
Hi everyone!
So I’m assuming you’re joining me from Chapter 2 of my story which talks about how I chose a boot camp and which resources I used to upskill. If not, please check out Chapter 2: The Uphill Climb to Upskill.
In this post (which is Chapter 3 of a multi-part series), I cover my experience with the job hunt process.
To kick this story off, I need to inform you that there’s never a right degree. You could have a PhD in AI and recruiters will still have you second guessing all your professional and academic decisions.
Data science interviews can feel like a special brand of hell.
Aside from the normal pain of job hunting and interviewing that you’d experience for non-technical roles, data science interviews are made particularly challenging by the following:
- Data science and machine learning roles are relatively new and people with the same title could be covering a wide span of topics, tools, techniques and problems that are constantly evolving and being innovated
- Data science and machine learning can be applied in a diverse set of use cases, from medical to real estate to games, from sales to marketing to product design to drug delivery
- Data science and machine learning concepts aren’t easy to understand– you do need to rigorously develop specific prerequisite knowledge beyond what the typical American college major usually provides (this is not a knock on American universities — I went to UC San Diego and had a great experience — mostly).
Great, so why care?
It’s important to call out what’s unique about a role as a data scientist (or machine learning practitioner) because the typical data science candidate’s experience can feel incredibly jarring and disconnected even between interviews for roles at the same company.
As a data scientist candidate, you can be expected to do some version of:
·1 -HR phone screen
·2- Hiring manager phone screen
·3- Technical screen (either sent or live) — Usually in a coding language of your choice
·4- Take-home case study (either performing exploratory analysis or building a model)
·5- On-site whiteboarding and/or panel presentation and/or team member interviews
And you can be expected to go through some of those stages multiple times.
One company might value coding skills, one company might value tool knowledge, another company might kick you from the pipeline because you couldn’t whiteboard out or live-code an implementation of the Xgboost algorithm from scratch (this happened).
And regardless of whether you would perform well on the job, a syntax issue on an otherwise great query means you’re not proceeding to the round (especially if the grading is automated).
And for data science roles at mid-tier companies, you could be competing against hundreds of candidates (especially for entry-level data science positions).
Engineering interviews seem to draw the closest comparisons in describing the data science interviewing circus but if you’re not familiar with how data science interviews or engineering interviews are structured, then keep reading to learn what I needed to do in order to eventually secure a job as a data scientist.
Chapter 3: The Hunt Begins
The brick walls are there for a reason. The brick walls are not there to keep us out. The brick walls are there to give us a chance to show how badly we want something. Because the brick walls are there to stop the people who don’t want it badly enough. They’re there to stop the other people. — Randy Pausch, The Last Lecture
Success is stumbling from failure to failure with no loss of enthusiasm– Winston Churchill
Who Are You and What Do You Want
When I think back to my job search, the work I undertook to get to my offer consisted of three stages:
(1) Defining my identity and crafting my story;
(2) Resolving knowledge and candidacy gaps; and
(3) Embracing the job hunt grind (including the ups and downs).
The most important factor in determining the success of a candidate’s job search is grit.
When I completed the Springboard Data Science bootcamp I was still experiencing doubt and internally I felt like a fraud (still do, working on it).
I was anxious whether I was “good enough” and I was constantly on edge about whether connecting and helping others would “take opportunities from me”, whether I could “compete and win”.
Anxiety about how I was perceived, insecurity about the lack of letters next to my name.
The constant comparisons were exhausting.
I didn’t feel confident in interviews and even though my resume would get great feedback (along with the two projects in my portfolio) I knew I needed some assistance in crafting a cohesive narrative of my professional and personal experience. I also needed some encouragement from data scientists that had fought the good fight (and won).
Months before starting Springboard I had stumbled across Kyle McKiou’s Data Science Dream Job mentoring program. I was hesitant about the potential ROI. But decided to connect with Kyle on LinkedIn anyway so I could follow his content and LinkedIn posts. At some point during Springboard I also signed up for one of Kyle’s webinars to get access to the Portfolio mini-course and was impressed with his down-to-earth explanations and easily digestible format.
After completing Springboard and feeling stressed about the impending job search it felt like the right time to sign up for the DSDJ program.
Stage 1: Defining My Identity and Crafting My Story
If you know the enemy and know yourself, you need not fear the result of a hundred battles. If you know yourself but not the enemy, for every victory gained you will also suffer a defeat. If you know neither the enemy nor yourself, you will succumb in every battle.― Sun Tzu, The Art of War
In the fields of observation chance favours only the prepared mind. — Louis Pasteur
Mental Mastery is introduced early in the DSDJ website’s curriculum because of the central importance of grit, capability and positive psychology to anyone’s success, whether personal or professional.
Some key resources that were recommended early included Angela Duckworth’s Grit, readings from Carol Dweck, and other leaders in positive psychology like Anders Ericcson.
Through a combination of Kyle’s talks, reading recommendations and module exercises included in the program, I began to notice significant shifts in my mindset.
Some of the most significant changes included:
- Being able to recognize constructive feedback as a sign that someone cares about where you’re headed and is trying to suggest ways to get you there faster and more effectively. Not a personal judgement or labeling of your worth.
- Being comfortable with failure and reframing failure as learning. In the course of challenging and pushing beyond your existing boundaries, you’re going to take some risks and some of them won’t pay off. And that’s okay. Rather than laying on the ground you pick yourself back up and adjust.
- Being able to see that my personal and professional experiences are additives to the right team, at the right company, at the right time. Just because my background was unconventional from an education and early professional experience standpoint didn’t mean it wasn’t valuables.
- Being able to accept that I didn’t “deserve” an amazing, creative and financially rewarding career, or at least anymore than anyone else around me. No one inherently “deserves” the best career nor is anyone inherently “undeserving”. Instead people “earn” their best lives through their choices and decisions every day. And this is a particularly tricky concept because rather than saying, “No one deserves to be happy”, my interpretation is closer to “you get what you put in”.
After reprogramming my mindset, the next step was crafting my story and positioning my unique value proposition to employers.
A lesson I learned early on from working at my first startup (which was focused on recruitment tech) was from a manager, who quite succinctly summarized what many recruiters and hiring managers are probably thinking: “No cares about your ASPIRATIONS”. Hard lesson to swallow but true. Another version is “It’s not about what the job brings you but what you bring to the job.”
At the beginning of a data science job search, the temptation is to try to fit everyone’s definition of a data scientist and boil the ocean.
While keyword-matching your skills and experiences to job postings is important, as well as building really cool projects for future “You”’s portfolio, even more important is owning the pieces of who you are NOW and incorporating those pieces in your narrative.
There’s a fabulous chapter in Meg Jay’s book “The Defining Decade: Why the 30’s are not the new 20’s” that includes a vignette where she’s gnashing her teeth trying to get her counseling client to commit to a career. He works as a mechanic in a bike shop and is unsatisfied with his current role but won’t commit to an alternative career because every other career out there feels off the shelf and not extraordinary. Jay is finally able to reach him when she compares his custom bike to a career; a unique amalgamation consisting of off-the-wall parts built up over time.
Another way to think about the value of a unique story is considering the perspective of a recruiter or hiring manager.
When you have 20–30 resumes gliding across your desk, it’s easy for candidates to start blending in a pool of words and formatting.
Degrees, certifications, skills, recommendations.
Do they become just another bullet point on your resume or do they become a supporting point in the arc of your narrative?
And professing yourself as “passionate about data science”, “interested in” or “pursuing data science” isn’t enough either for your elevator pitch.
Hypothetically, everyone looking towards a career in data science is passionate about…data science. Your story needs to be far more unique and relevant and it needs to come from owning the experiences and skills you’ve already developed that can additional value.
Another way to think about your unique value proposition is thinking about your “superpower”. A “superpower” in this case is any skill that you feel is your competitive advantage. It can be technical but it doesn’t have to be. For all my job searches, the superpower I advertised was my expertise leveraging analysis in various business domains like product analytics, sales analytics, marketing analytics and growth. Specifically my ability to serve as a “data translator” (see these articles from McKinsey and Forbes) was my unique value proposition.
Once I owned my story, non-traditional by most standards for a data scientist, I utilized various DSDJ resources to put pen to paper and populate my resume and linkedin.
Stage 2: Resolving knowledge gaps
Everything you want is on the other side of fear.–- Jack Canfield
Real education enhances the dignity of a human being and increases his or her respect. If only the real sense of education could be realized by each individual and carried forward in every fields of human activity, the world will be so much a better place to live in. — A. P. J. Abdul Kalam
When all is said and done, however, I still needed to systematically resolve remaining gaps in my knowledge and skills.
The two strategies I used were:
(1) Creating a study plan around my weak areas and
(2) Taking on side projects for friends needing data science expertise.
Strategy 1: Creating a Study Plan
After taking a month to relax and catch-up with friends and family, I started writing up a study plan to structure my next 6–9 months of job applying and interviewing. The first draft involved taking every month to focus on the following topics:
- Data Cleaning + EDA (Visual Analysis)
- Statistics/Probability Theory/Experimentation
- Machine Learning 101
- Specialized Topics in ML
- Software Engineering Best Practices
- Specialized Topics in Math
I would use a combination of passive studying resources (videos, podcasts), mid-active studying (technical books, code review), and active studying (coding tutorials, online mini-courses) to focus on artifact creation (projects, code, notebooks, blog posts) as well as reworking my previous projects.
However like all best laid plans I had to be a bit more agile and change my focus throughout the job search.
Initially I had this really complicated learning plan that focused on stats for two months, then coding for two months etc. But the feedback I got in some of the early interviews was I needed more experience running and calculating A/B/n tests. After squeezing myself into some A/B/n testing projects at work and developing my understanding of experimentation and analysis, I would go into interviews and do well on the stats questions and then die on algorithms questions.
Rinse and repeat.
My take-away from the experience of creating a study plan and then going through a rotating cycle of rejections was that I needed to iterate through the different topics in a much shorter cycle.
The entire data science interview process at some point begins to feel like a fast paced version of whack-a-mole.
Different companies have different needs at different times and no single program or resource will adequately cover all the skills needed for the job.
Strategies and resources I used during this period include:
- Practicing on whiteboard wallpaper that I stuck all over my apartment
- Taking architecture grid notebooks and writing out example code blocks & notes while going through coding tutorials
- Practicing SQL and Python questions on HackerRank, LeetCode, and CodeWars
- Completing at least 1–2 DataCamp courses a week and utilizing relevant code snippets in personal projects or posting to team wiki
- Watching PyCon videos when possibleStrategy 2: Taking on Side Projects and Learning to Think Like a Business/Product Owner
- Purchasing used O’Reilly books (Practical Statistics for Data Scientists: 50 Essential Concepts by Bruce, Data Science from Scratch: First Principles with Python by Grus, Python for Data Analysis: Data Wrangling with Pandas, NumPy, and IPython by McKinney, Hands-On Machine Learning with Scikit-Learn and TensorFlow: Concepts, Tools, and Techniques to Build Intelligent Systems by Géron) and reading through them multiple-times on the commute, then going home and practicing the queries using Colab or Jupyter notebooks.
Strategy 2: Side Projects
During this period I also started helping out with side projects, from fitness to real-estate. My family didn’t understand why I would do “free work” after paying for a boot camp.
While I can’t say much about my friend’s projects right now, I can talk about the rationale behind why someone might want to consider contributing to other people’s projects, even if they’re not open-source.
The main reason why I spent some nights and weekends working on other people’s projects is the ability to get hands-on experience in areas like:
(1) Creating production quality code;
(2) Formulating prediction problems from scratch using challenging and disparate data;
(3) Model deployment;
(4) Feature engineering in a production-like setting.
The kind of experience and upskilling that comes from being an individual (and sole) contributor at an early stage start-up is a different experience from being part of a team on a mid- or late-stage company.
I initially just saw it as an opportunity to gain valuable experience and work with smart and interesting people while continuing my main job search for a data scientist role.
What happened was the minute I took co-ownership of the product, everything I‘d learned about product analytics, data science, and growth marketing started to click into place.
Suddenly the seemingly endless and opaque list of requirements for data scientists (like experience with proper statistical testing and experimental design, agile model deployment, good testing and debugging, end-to-end visibility of the product) started making sense, especially in the context of being a data science product owner.
Even if you’re not working on a side hustle, you can learn to adopt the same mindset by imagining that you’ve just been hired for a job as the sol data scientist for a start-up.
When you’re creating your own consumer machine learning product that needs to successfully ship and meet customer demand such that significant market share is captured in a short period of time, all your inputs and decisions matter, especially given you’re trying to pay yourself in the future.
After working on side projects I determined that for my next role I wanted to become the kind of data scientist that is:
(1) Creating internal data products that power customer-facing apps;
(2) Performing experimentation, segmentation and causal inference analysis;
(3) Incorporating machine learning into strategy, forecasting, and planning;
(4) Constantly learning & pushing the boundaries of my knowledge & experience with advanced machine learning topics, like reinforcement learning, deep learning, computer vision, autonomous vehicles
Stage 3: Embracing the Highs and Lows of the Job Search
Never stop fighting until you arrive at your destined place — that is, the unique you. Have an aim in life, continuously acquire knowledge, work hard, and have perseverance to realize the great life. — A. P. J. Abdul Kalam
I must not fear.
Fear is the mind-killer.
Fear is the little-death that brings total obliteration.
I will face my fear.
I will permit it to pass over me and through me.
And when it has gone past I will turn the inner eye to see its path.
Where the fear has gone there will be nothing.
Only I will remain. — Bene Gesserit, Frank Herbert
Putting in the Emotional Work
Sometimes the hardest part of the job search, whether data science or not, is the emotional work. It’s easy to point to all the reasons why you can’t succeed or won’t succeed. But what if the problem isn’t the skills, experience, background…what if the problem is YOU?
What if the problem wasn’t with all the things you couldn’t control but all the things you could?
It can be an incredibly uncomfortable experience looking in the mirror and being honest with our shortcomings.
In my case there was security in not taking accountability and ownership of my success. There was security in saying “One day” and then trying to control the job hunt situation by planning, planning and more planning. Planning and never doing.
During an office hour call with one of the DSDJ mentors (who I’m now happy to call a colleague), I showed off my elaborate study plan and talked about how I was going to target applying for jobs “6–9 months in the future”. I promptly got my hand (digitally) slapped and was challenged on the timeline. “Why WOULDN’T you try apply now to at least five jobs?”. The question was fair and Harpreet encouraged me to suck it up, do the work, and apply to at least five jobs on LinkedIn. Worst case scenario I get rejected, adjust my resume and study based on the original timeline.
Committing to achieving the outcome
After a month of LinkedIn applying I started getting some bites from recruiters for data scientist roles heavily focused around decision science.
Regardless of advice I’ve seen to tailor portfolios around specific data sets and problems in specific industries, I received call backs from every type of company in every type of industry possible.
At some point every company needs to sell and my background working with sales and product analytics meant I had a valuable skill set that wasn’t industry specific.
Of the 5–15 applications I sent out every day, I’d get positive responses from 15–20% of the roles (the remainder being auto generated rejections).
Over the month I would go through a very similar process of: recruiter call, team member screen, technical screen via CoderPad, hiring manager call, take home study, onsite interview of 4–12 interviewers (with some rounds being whiteboarding sessions and others being walk-throughs of projects I did, along with behavioral questions).
Whenever I didn’t make it to the next round I’d try to get as much feedback as possible and incorporate the feedback into changes in my study plan.
After the second month of purely focusing on interviews I realized I needed more time in my schedule to productively break from the constant grind and work on interesting data science problems, even if they were just personal projects or snippets.
After a skype session with one of my best friends Veena where she talked about getting up at ~ 3:30–4 am to compose music for two hours before going into work as an engineering manager, I was inspired to do something similar.
My new schedule consisted of:
- Waking up 2–2.5 hours earlier every morning (~ 5:30–6:00 am) to code and study for two hours;
- Heading into work around 8am and using the commute to read or watch data science related videos;
- Use lunch time/midday to do more coding and read through tutorials between 12pm — 1:30pm;
- Leave work around 4:30pm to get to the gym or the boxing club for 2–3 hours of training;
- After heading home for dinner and a shower, turning off all electronics at 9pm ;
- Blocking off the last 1–2 hours of the night to improve my sleep hygiene and relax.
I was still using weekends to work on side projects with friends and take of chores, as well as spend time with family.
By front loading the most important tasks in the morning when no one was awake and I had the most energy, my mood improved drastically and I was able to produce more work.
By the end of month three I was getting into a cadence of study, apply, interview, more apply and more interview. But the offer I wanted always seemed just out of reach and I was losing a little bit of steam.
Was it all a waste? Was all the effort worth it? Did I make a mistake? Or several, kind of expensive mistakes?
Thanks for reading this far!
If you’re interested in learning more about my experience with the data science job search process, I write about getting the offer and next steps in Chapter 4. I also answer some common questions that come up frequently on LinkedIn and in-person.
Interested in following along with my Data Science journey? One of the easiest ways is by connecting through LinkedIn. Feel free to send me an invite and let me know what you thought about my series! And if you really love what I wrote, consider buying me a coffee at www.buymeacoffee.com/mmbazel & keep me writing! ☕
You can also find me on GitHub or at Kyle’s Data Science Dream Job mentoring program if you’re interested in being mentored by me.
If you’re curious about the series or want to jump between parts, feel free to check out the rest of the series:
● [You Are Here✔️ ➡️] Chapter 3: The Hunt Begins (aka “How I Got a Job as a Data Scientist”) — This section covers my experience job hunting, from applications to the onsite interview. Key Takeaway: Swing for the fences and dig deep! 🏏
