Career Update May 2021

Introduction

In this post (September 2020) it was decided to move to a Data Science job, after which I can comeback to figuring out what I needed to do to get into EA. Basically Data Science is my backup and I would start my journey in here, while I work on the side to move towards EA.

Am I already good enough?

As of September 2020, I had already spent about 800 hours on doing the Coursera Capstone Series for Data Science and as part of it I have some “projects” under my belt. In addition, I have also been “scripting” in different contexts for over 5 years (at universities) and I have 1 year experience in contributing python code to an open source software.

It is “well known” that people from various branches of education retrain and get into Data Science. In the semiconductor company I work in, I am aware of people who moved into data science from Physics (back in 2018). Another example, is a PhD in Neuroscience who moved to a Data Science Job with 6 months of training, she claims here (Jan 2020). She barely knew how to code when she started of. However it appears she has years of experience in design of experiments, statistics etc.

I personally know a few people from the university I did my Master’s in. A guy with an embedded systems Master who did a 9 month thesis on Deep learning, and a girl from Energy Engineering who did her internship in shell in “data management”, got into roles titled “Data Scientist”. (It was back in 2017 when they got the job.)

Looking at the experience of other people (who did a master’s like me from Netherlands and at the same institution) maybe I was already there, and applying might just end up working out.

Applying

With the help of an Apple Data Scientist (who looks at a 50-100 CV’s a year for hiring), I made a new CV and applied to 25 companies. But I didn’t get a single call for Junior roles even (“Jr. Data Scientist”). I applied to exactly 25 companies that I found on LinkedIn Jobs (neglecting the experience or language requirements listed¹).

Upon requesting feedback, one HR from “bol.com” (company in Netherlands) told me that I had a “mechanical degree” (implying that it is not “math”), and that there were 100’s of people who applied and “many of them” had experience (unlike me).

A “team lead” from another company seemed interested in me when I called him up. Even though I was not a fit for the “data science consulting role” that he had now, he put me in touch with another branch. Unfortunately, they didn’t have any vacancies. He also informed me that “doing Kaggle” is a “right direction” for me to “get experience” and that he also worked on Kaggle competitions with his colleagues.

Some people responded to my applications asking me if I knew Dutch. Unfortunately, I am not good enough in Dutch for them.

Many vacancies wanted me to be able to use decision trees and other ML techniques (amidst other things), which I didn’t have any experience in. As a result these techniques were not stated in my CV. This could have also been an issue.

At this point I could continue to apply to more companies (it was also nearing December) or I could work on myself for the coming months. I had no belief that sending more applications would get me anywhere, especially after the responses I was receiving for the current applications. And here I thought Kaggle was the only way to “gain experience”, gain credentials to write on CV and then, job!

“Working on myself”

I “fixed” my CV with clarifying that Mech master’s is math intensive (in Linear Algebra and Calculus). I did a course on Machine Learning and then went on with doing Kaggle for 2-3 months.

I worked on legacy competitions so that I could learn from the best solutions. I would try a solution myself and learn from the greats what the better solution was, and then implement it. After grinding for about 3 months I ended up with top 10% scores and top 20% scores in 2 sets of competitions in applied machine learning. Yes, I did get help from the actual solutions for one of the competitions, but there was still a lot of work to do and I learned a “lot” working on it. This part about me not owning my solutions a 100%, barely mattered in the end (the next section has the details).

I made a blog and wrote my experiences in Deep Learning and Machine Learning. I highlighted in my CV my achievement of “top 10% solutions in Kaggle”. Boasting such a “powerful” credential it was very likely to grab the eyes of recruiters. Well!

Apply again

This time I applied to 60+ vacancies (neglecting the experience or language requirements listed). I stopped counting after a point. I expect the actual number to be in the order of 80-100 applications. I applied to full-time vacancies I mostly saw on LinkedIn Jobs, 2 through referral and a 2-5 internships too. In addition I started speaking to several Managers (15-20) within my company to look for jobs inside.

No one seemed to care about Kaggle :(. Half way through the application process (30-40 applications later) I realized no one is interested in hiring me as a Data Scientist. 0 interview calls. I got a call from an HR who asked me what this Kaggle was and in the end realized that it was not “real work experience” and didn’t come back to me after.

Lowering my expectations I started applying to titles such as Data Analyst and Business Analyst. I tried a few things over the course of applying to these companies, such as sending the same CV substituting “data science” with “data analyst”, removing image processing credentials, putting focus on “SQL, python, tableau”, etc. This time I got 2 calls for the next stage from other companies and 2 calls from within my company.

I got 1 call from a company called Picnic, but they had such high standards for entry where I didn’t manage to clear their “IQ test”. I ended up with a 47 percentile score while they needed a 70 percentile score. I got another call from a second company who said they had a “good interview” but they thought I was too junior and needed someone who could “already run from the get-go”.

Within my company I managed to secure 2 jobs where “data analysis” is involved. I am expected to “deliver insights from customer data”, while interacting with a dozen or more “domain experts” and tell a concise story in the end. It seems to focus on “diagnostic” and “prescriptive” “data analysis” and not on the “predictive” or “descriptive” “data analysis”.

But what happened? Why didn’t a single soul from outside my company want me?

Post-mortem

I still don’t know whats wrong or missing with me and what I could do to improve my chances. Rejection emails were quite useless with responses like, “there are other candidates who fit the requirements better than me”. Mailing people back for feedback seldom resulted in replies. When I called them, they simply didn’t pickup. Repeatedly following up was taking the whole day with barely any progress. I spent most of a day calling just 5-6 companies and getting in touch with about 1 in the end. It was taking the entire day to focus on this and make the calls, prep before the call etc. Very different experience than what I had in 2016. Everyone would answer my call back then. EVERYONE!

Many jobs (35% of applications as per my notes), I applied to wanted work experience of >1 year as a Data Analyst.

I think I don’t fit in the Data Analyst pool (SQL, visualization, python, cloud exp), or the Data scientist (machine learning in addition to Data Analyst skills) pool. For example, for a Junior Data Analyst position in ABN AMRO which basically needed python, SQL and visualization, I got the feedback that I was rejected because people like me will be bored in 6 months, + there were 100s of applications and that I might be suited for “Advanced Analytics”. And when I apply to Data Science positions which require Machine Learning (“advanced Analytics”), I barely get an interview.

It appears that knowing Dutch would have helped. Quite a “few” companies replied to my initial application, checking with me if I knew Dutch, and consequently rejected me because I didn’t know Dutch. Around 25% of the companies at least were explicitly mentioning that they needed dutch (as far as I made notes on this). There were also companies that didn’t mention anything in the job details but said in the rejection letter that they require fluency in dutch. I even tried traineeships geared to train you and send you to a proper job in data. But they also rejected me on grounds of not knowing Dutch.

Within my company, the feedback I got was varied. For data science positions, either people didn’t have a vacancy or they wanted someone with “real experience” (or a “real education”) or in some extreme cases, ONLY PhDs. WTF!

What did other people do?

I looked at one of the “popular” bootcamps here in Netherlands (Ironhack). This bootcamp boasts “reasonable” statistics after 2020 January: 67% people got “what they wanted” (job, internship, knowledge) within 3 months and 76% got “what they wanted” in 6 months time.

I found a lot of profiles from the bootcamp and focused on people who graduated after May 2020. I contacted many of them and some of them were kind enough to give me really detailed responses. From talking to these people and in many cases just looking at their LinkedIn profile, I have the following data on what people are doing, how long it takes them to get a job etc.:

1. Some people who had a background in Translation are just starting internships after 7months, in a small company (9 employees).
2. Some people with Media, or with chemical engineering background but no real experience, managed to find “basic” (optimization of web analytics, using google analytics) data analyst jobs after 5-7 months, in an small company (18 employees).
3. Some people who are from an engineering background very similar to mine started their internship (in spain) after 1-2 months of end of bootcamp. And then followed it up with an actual job as a “data engineer” in another 1-2 months in a startup (27 employees).
4. Some people who have an MBA were able to manage jobs within one and 6 months of end of bootcamp. [26 employees]
5. Some people with marketing experience are finding jobs in the same domain within 2 months with “multiple offers” and a small number of applications (30). This is also a small company (67 employees).
6. Some people with a PhD in neuroscience, are able to get a job after 4 months of bootcamp. This guy made 100+ applications with only 3 being converted to interviews, eventually landing a job in a mid-sized company (1900 employees).

* internship in spain and then job in Netherlands as “data engineer”. ** 30 applications with “multiple offers”, didn’t qualify “IQ tests” and had the luxury to say no to companies that didn’t interest her. *** Applied to 100+ companies, failed “IQ tests” and made it to 3 interviews.

When an STM mentioned on Jan 5 2021 that maybe I should try to find more people “similar” to me, I thought I couldn’t find more examples than the ones I had in the beginning (I was wrong). These examples would have been nice to know. They give insight into where to apply, how to find those companies, what people in the pandemic are doing to get jobs, if it is even possible to get an internship etc.

What I didn’t do that other people did

I didn’t do this bootcamp, but I don’t think that was the reason why I didn’t find a job. Knowledge wise I think I have the skills taught in the bootcamp. The syllabus looks the same as what I have done in the past. People who have completed the bootcamp also think that I might not need this. The bootcamp is targeted for people with 0 coding experience while I am not really a beginner. I also compared the type of output these people are generating vs mine. For example, I compared my tableau dashboards to theirs and it looks just as “simple”.

In addition to courses, the bootcamp has a career week where they give you tips on your CV and “help you with your job search”. When I checked with people from the bootcamp, they suggested it was “basic” and that I could afford skipping it if I have applied before.

I can however still learn from what people actually did to get jobs such as additional projects done, how their CV looks and types of companies they applied to. People don’t seem to have worked on additional projects besides those in the bootcamp. Their CV looks way “clearer” than mine and they stuck to 1 page instead of my wordy 2 page CV. And most of them applied to “small companies/startups”. I didn’t know how to find such companies until I asked some of them and they shared links. I will replicate these the next time I am applying.

I applied for less than 2 months while most of them seem to have applied for more than that. I did maybe 80-100 applications (half of it was for “data science” roles) while some people did more than 100 applications for data analysis roles alone. Regular applying and rejections was getting very hard for me to bear amidst the total lack of feedback. And I just wanted to settle for anything after those 2 months of disappointment.

With this new information, I could modify my CV, target startups exclusively and see where that gets me over another 3-6 months. Should I though? Don’t I have a job already in “Data Analysis”?

Is this the “right” data analysis for me?

I managed 2 similar “data analysis” jobs within my current company. But I don’t know if this is the “right data analysis” role for me. It feels like this job might be missing skills that “most” jobs outside the company require. Out of 27 applications I made for the role of Data analysis (and have recorded data on), at least 15 of them wanted SQL, at least 12 of them wanted python/R and at least 10 of them wanted Visualization skills in Tableau or other similar software. None of which I will be doing in this job. Everything is done in MATLAB.

As a result, I could potentially have a “growth” which lags by “X” years from that of a person who does data analysis outside my company (say by virtue of using SQL, python and Visualizations). And this “X” might be important to know quantitatively as it drives the answer to the question: Should I continue to invest time in finding the “right job”, or Should I continue with the “data analysis” role within my company?

The Final Trade-off

These are some of the options I have, and using lifetime calculations for each scenario it is possible to see which is better:

1. I could look for another 6 months to 1 year to get a job or internship in say “e-commerce”, “web-analytics” etc., and end up as a data analyst. And then start with focusing on EA. Essentially this is delaying the EA path by 1 year. (Lifetime calculations here: “Opportunity Costs 3”)

2. I can continue the Data Analyst job that I have now and move on to focusing on the important questions related to EA. I will even assume in this case that compared to option 1, this plan will set me back by 2 years in case I don’t “make it” in EA. This way I make a conservative estimate. (Lifetime Calculations here: “Opportunity Costs 4”)

3. Do a Post-master’s program–recommended by a manager in my company who didn’t want to hire me now–for 2 years in Netherlands and then find a Data Science job and then start with focusing on EA.

   It will be 3 years from now, before I finish it (Age 33-34) and get a job. It’s already costly for delaying the EA path by 1 year (as shown below with Lifetime calculations for cases 1 and 2), let alone by 3 years. In addition starting to learn new skills after 30 is going to be “hard” [citation needed]. So the sooner we start in the EA direction the better.

Lifetime calculations for the above 2 out of 3 cases are shown:

It seems that continuing with the Data Analyst job that I have now in my current company and focusing on EA right away, is the better option.

Conclusion

Am happy that I am in some sort of a “data analysis” position, and that I get “some experience” while my daytime job is not fully a waste of time. As the calculations seem to suggest the delay in finding a “real Data analysis job” might not be detrimental. So for now, am going to get to the bottom of the questions I left hanging in the last post such as, “what I need to learn/do to get into EA”? “Can I get empirical evidence of lifetime potential of an EA job”? etc.

In any case, I think it is a good idea to re-evaluate the situation in 6 months. I will have more idea about my job in a few months. I will get to talk to several people and get examples of where this path has taken others (#examples). Meanwhile I will also “network a bit” by joining “meetups” to meet other similar data science folk.

Timeline TL;Dr

September 15 –> wrap up post and start prepping computer for Machine Learning course.

November –> prepare CV, get feedback and apply to 25 companies and work on the side on Machine Learning course.

December-February –> Kaggle competitions

March-April –> Apply to 80-100 jobs

May –> Write the blog post, find and discuss with alum of Ironhack

Stats

3500 words and 34 hours.

Appendix

Data Science skills and how often they required by companies

python –> 16/30

SQL –> >10

ML,decision tree, RF, regression, classification etc… –> 15

DL, CNN, NN –> 2

viz –> 5

Statistics –> at least 4

pyspark –> 3

hadoop –> 3

kubernettes –> 1

julia –>2

tensorflow –> 2

pytorch –> 1

azure –> 3

aws –> 4

gcp –> 1

ETL –> 0

> 1 year –> 13

Data Analytics Skills and how often they were required by companies

Entire list: SQL, excel, data visualization–tableau (BI softwares), java, spark hadoop, pyspark, aws usage, ETL, azure, kubernettes hadoop docker, R

SQL –> 15/27

BI –> 10 at least

python,R –> 10 at least

docker –> 1 at least

kubernettes –> 2 at least

hadoop –> 3 at least

spark –> 1

java –> 2 at least

>1 yoe –> 8 at least

ETL (data engineering) –> 2 at least

More than 100 apps according to LinkedIn –> 10-15 apps

jda –> 3 at least

da –> 20 at least

ba –> 2 at least

pyspark –> 0 Da positions

azure –> 2

GCP –> 0

AWS –> 1

According to the Apple Data Scientist I mentioned in this post, the following are the most important skills for “data analytics” jobs: writing queries (SQL), cleaning, manipulation, adding features, making data products, insights and presentation, correlation analysis, hypothesis testing, create data impact, statistics

Related work done in other non-published files

2020-06-11-data-science-what-to-do.markdown

2021-05-07-Data-Science-so-far.markdown

Footnotes