A year as a Data Scientist right after college: An honest review .
with a pinch of salt.
I would like to give a brief background about me before going any further to help you understand where I am coming from.
Since my school days I enjoyed doing mathematics and learned to do basic programming. Later on, I studied Electrical Engineering at IIT Bombay and worked with a Harvard University professor for my year-long Master’s project in Scientific Computing. After graduation, I joined a Consulting firm specializing in AI, where I worked on data science projects for clients in India and Europe.
I must say I am still in early stage of learning. I do make mistakes sometimes and wonder: what have I learned during all these years?
The more I learn, the less I realize I know. -Socrates
Observations/Comments (Unvalidated Hypothesis)
As a data scientist, you don’t control data, you give it a shape so that it can better express itself.
1. Data Science can be really fun if…
Data science is a rare job where you get to do all of the cool stuff together: mathematics, coding, and research. A job where you can read a research paper in the morning, write down the algorithm in afternoon, and code it up in the evening. It is really fun! A few days back when I visited my alma mater, a junior asked me: how would I describe my job experience?, to which instantaneous reply came out of mouth:
It is like getting paid for doing assignments!
But here’s the catch: you have to do “only some” of math, coding, and research. You can neither leave any of it nor afford to go in depth in any of it. In some cases, you might not have to read any research paper and you can directly use code libraries and start implementing. And given the time constraint, the only part in which you will ever get the chance to go in depth is the data preparation (and the presentations to be made, if any).
If you are a person who loves to code and want to do it as much as possible, the math will bite off your head. You won’t get the chance to showcase your Ninja coding skills. As a code-lover turned Data Scientist you have only two options: either math will conquer you or you will conquer it. If you are a math lover who likes to solve complex problem, the data preparation and monotonous programming will bore you to death. The analogy I like is:
Imagine how will it feel if someone gives you a stapler and sends you into a room full of papers to be ordered and stapled? That’s how the data cleaning process feels like.
Sometimes you may feel that if someone else could do the data preparation work and you can just build the model around it, but unfortunately or fortunately it doesn’t work this way.
One may argue that the kind of stuff that you don’t like only occupies a small part of the job, but the thing is, it tends to have a disproportionate effect on your work (read the Pareto principle).
The way to sustain in data science is to strike the right balance between coding and mathematics/research. And most importantly, appreciate the different aspects of your job while maintaining the focus on getting the results.
Also, data science being an evolving field and with no definite guide to success, requires a lot of hard work, continuous learning and most importantly unlearning (you don’t know when the “best” will become just the “good”).
Pros: data science lets you work with a lot of interesting stuff at once: code, math, research and sometimes presentations too
Cons: data science may not allow you to go in-depth in any of its individual aspects
2. Data Science is more of “business” than “science”.
“What’s in a name?” asked a wise person. Two replies echoed:
“Nothing.”, said the philosopher
“Everything.”, said the marketers
Data Scientists in the industry are unlike scientists working in say, CERN or ISRO. What Data Scientists are “actually” doing is applying scientific techniques to create value for the business. Data Scientists job is not simply to find the best solution but to find the solution which is easily interpretable and sell-able.
In data science, when encountered with competing hypotheses, business logic often trumps the scientific logic.
The knowledge of Artificial Intelligence, especially its subfield Machine Learning is enough to get you started but not enough to get you ahead.
To excel in data science projects, “science” is not sufficient, you also need to understand economics and marketing to get your work through and create value for the business.
3. Data Scientist’s work is impactful but like any “other” job.
As a Data Scientist you have “potential” to make huge “impact” but so you do have as a teacher, software engineer, journalist, or salesman.
Data science is unarguably changing our world but the areas where as a Data Scientist you can play a crucial role & “influence” are rare.
Data science being circulated as an enigma to “change” the world on the Internet is not always true.
Moreover, the term “impact” is highly subjective and have many important aspects to it like:
how much monetary value does it have/create
who will it benefit the most
how unique and important your role is in it
to what extent is it self-perpetuating
The kind of “impact” that you want to have, one the Internet is talking about, and what the business can provide you the opportunity for, can be very different.
If you are a smart and ambitious person getting into data science to make an “impact”, make sure you are right “fit” for the field and your meaning of impact is aligned with the impact that the business can provide you.
4. Economics today dictate data science.
As of today, Data Scientist is a “hot” job and it is not uncommon for multiple recruiters to contact you regularly to see if you wish to switch. Data Scientists are given a lot of respect within the organization and are paid handsomely.
I don’t recommend that you become a Data Scientist if money is your primary motivation.
If you are doing data science only for money, mind you sooner or later the same economics will make the high salaries fade away. It is only your interest in the field that will keep you going.
5. Data Scientist’s experience as a stepping stone
Working as a Data Scientist can help you build a strong foundation for the data-driven world and more realistically assess limitations and power of AI-related technologies.
Even if you choose to step away from the hands-on work, the experience of Data Scientist could be valuable in your future endeavors. Some of them are:
Academic Research(Masters/PhD/Post Doc/Independent): A Data Scientist’s considerable amount of time is spent doing experimentation, reading scientific papers and debating ideas with their colleagues. All these activities can prove to be valuable assets for doing research.
Startups: Of course, you can work in a tech role in an AI startup or start your own after working as a Data Scientist. But, you can also take up non-tech roles in an AI startup like sales, business planning or finance. Your knowledge can help you better understand and sell the product to customers, estimate project costs and requirements, and most importantly not feel like an outsider. There can be various reasons why you may want to work in a non-tech role after gaining experience as a Data Scientist, one could be you mattered to look over the number of people who regularly code in the Executive team of any successful tech company.
Government and Global Organizations: Governments are becoming more and more concerned about the impact of AI systems on society and how to best utilize them for sustainable development. You can use your Data Scientist’s experience to work in roles related toAI policy, where a sophisticated understanding of AI is required to make policies and regulations that are best for the society.
Note: These are my personal views and you may or may not agree with them. My intention was not to write about what an ideal Data Scientist work looks like, or how “should” data science be done, or even to describe who is a Data Scientist. These were few important pointers, I wanted to share, that I learned while working as a “Data Scientist” for a year right after college.
Your experience can be very different from the one I had described above. I encourage you to share your experience in comments below or write separately about it. It would be great help for the people who want to dig deeper into data science.
In case you liked the article, do follow meAbhishek Parbhakarfor more articles related to AI, philosophy and economics