Here’s how to develop the qualities that will help your application stand out
Landing a job in data science can sometimes feel like throwing darts at a board blindfolded — sometimes you’re way off the board in your application and don’t even get an initial interview, sometimes you land close to the bullseye only to not fit the candidate mold perfectly in the technical interview, and sometimes you hit the bullseye and land a job offer.
Sometimes there’s no rhyme or reason for how you did when throwing your dart blindfolded.
This means that it can be equal parts unnerving, frustrating, and crazy when you go to throw your hat in for a job that you can see has hundreds of applicants. You know the type of job I’m talking about — some basic entry-level position for a mid-level company with hundreds of data scientists clamoring for their shot to be discovered. This may be the type of job that you apply for just because it’s there with no real belief that you’ll get it.
However, what if you knew what recruiters were looking for? What if you knew that you could improve your dart-throwing ability by just presenting recruiters with what they’re after? What if you knew that of the hundreds of possible candidates, you and only a handful of others are the real competitors who hold all of the attributes that recruiters are dying for on their teams?
You’d probably go into the application process with a lot more confidence, wouldn’t you?
Over the last several months, I’ve paid attention to recruiter posts on LinkedIn and Reddit that have outlined their struggles in finding ideal data science candidates. Recruiters from varying backgrounds, methodologies, and company sizes have reported the same consistent problems with candidates. From these descriptions, I’ve been able to put together the ideal attributes that data science candidates should have according to the discrepancies noticed during the hiring process, and have coupled these attributes with advice from my own personal experience that you can use to achieve these attributes. Your job search process will no longer involve blindfolded dart-throwing — you just need to focus on developing these eight attributes to help you become the well-rounded data science candidate recruiters are looking for.
It’s estimated that 40% of job candidates lie on their resumes and three out of four recruiters have caught someone lying on their resumes. These are pretty substantial numbers when you consider that hundreds of candidates may be applying for the same data science role at one time. The most common lies found on a resume include education, the time spent with a previous employer, and past salaries. Further lies can be found where candidates have exaggerated impact numbers (such as increasing a company’s data pipeline efficiency by 75%), used inaccurate job titles, lied about their technical abilities, claimed programming (and spoken) language fluency, or even used a fake address.
Therefore, the key here is to always provide recruiters with a recent, updated, and accurate resume that is designed with the job description in mind, represents your current qualifications, and is free from lies or other information that will keep you from getting the job. This may sound easy, but when you’re applying for hundreds of jobs it can sometimes be easy to fall into the trap of sending the same resume to all the companies and hoping one sticks.
Each time I apply for a job, I craft a unique resume keeping the parts that are relevant to the job and rewording sections to make them more applicable to the company. I rarely craft a cover letter unless it’s required to apply, and instead, focus on developing the best one page of information I possibly can. I also focus on a non-graphical resume to ensure that it doesn’t run into parsing issues during the application process. With recruiters only giving a few seconds to your resume, I try to keep it short, sweet, and focused on my ability to produce impact rather than the certificates I’ve amassed or courses I’ve completed.
How to improve this
It’s vital to take the time to produce a tailored resume for each job. This not only shows that you have an understanding of the job requirements but it also shows the recruiter that you have attention to detail and genuinely want the job. Your resume should include all relevant job experience, your past education, certificates you’ve completed, and a list of personal projects that are relevant to the job.
I became a favorite to have on a team for group projects when I was in university studying software development because I was a ringer for presentations who could communicate the value of our project to a wide range of audiences.
I got my first job in tech because I had superior written communication skills that allowed me to explain technical concepts to non-technical clients and potential clients and why they would benefit their company.
Throughout my time working in tech, I was on the front lines of communicating with clients because they felt comfortable talking with someone who could go between them and the technical team to ensure that business problems were being correctly translated into technical cases and vice versa.
In short, I got to where I was in tech by having communication skills.
Data science recruiters are particularly looking for candidates who know what they’re doing inside and out and can answer a variety of questions in interviews that are applicable to daily work. For example, recruiters are looking for candidates who can answer questions in their technical interviews about what they’re doing that go far beyond “Oh, I don’t know, it’s probably in the documentation”.
How to improve this
Unfortunately, communication skills aren’t focused on or taught in traditional data science learning pathways, which means that candidates who succeed in this area have these skills naturally. While communications courses are often taught as part of the program in university or college, one or two courses over 2–4 years generally aren’t enough to make you a competent or comfortable communicator.
Therefore, the only way to develop better communication skills is to practice. My favorite way to practice is to tell my very begrudging dog about everything that I’m doing, from why I’m writing the code the way I am to why I’m using this mathematical construct instead of another. Treating my dog like a recruiter sitting in on a technical interview is not only a great way to practice communicating what I’m working on, but it also gets you in the rhythm of talking while working. I went to school with lots of people who can only code in perfect silence. Even taking breaks to describe what they’re doing would throw them off because they had never practiced.
While some companies may not mind whether or not you can describe what you’re doing while you’re doing it, they’ll certainly want you to be able to give a full rundown at the end, as well as be able to descriptively answer any questions they throw at you beyond, “Oh, I’m sure you can find that in the documentation”.
Because the tech industry is on a sliding scale of requiring and not requiring credentials for a job, you must have the skills to support the job, regardless of whether your credentials are what they should be. This is often the reason why you’ll see hundreds of applications for a single data science job — because the field doesn’t always require credentials, anyone can apply to the job.
This results in many applicants not having the skills to support the job despite applying anyway. Because let’s be honest — how many of us have applied for jobs saying we’re Excel gurus without knowing much beyond =SUM()?
Recruiters seem to be coming across many instances of people not having the right skills for the job (see point 1 above about lying on your resume) which means that while a job may have hundreds of applicants, there may only be 10 or so actual hopefuls.
Luckily, the skills required for the job are always well-outlined in the job description which can give you a good starting point for your studies.
How to improve this
Given that you’re reading this article, I can assume that you’re either a trained data scientist trying to find out what recruiters are looking for, or you’re someone studying data science who’s trying to find out what recruiters are looking for so you can tailor your learning experience. From these assumptions, I can already tell that you probably have the skills for the job that you’re applying to.
The real issue that arises here is with people applying to data science jobs who have no real skills in the area.
Instead of being this person who wastes the recruiter’s time, just read the job description and ensure that you have the skills required. One of my favorite activities is reading job descriptions to see what companies are looking for at a given time. If I don’t have all the skills they’re looking for, I make a list and then begin learning some of them. I do this for all the job ads that I look at so at the end of the learning process, I’ll be well positioned to apply for any data science job that comes across my path.
Not sure where to begin? Try these resources to get started:
I’ve said it before and I’ll say it again: recruiters are only looking for candidates who are capable of producing an impact on the company the moment they join the team. One of the key determinants of this is a candidate’s track record of quality.
There’s not much more to say here beyond the fact that recruiters are looking for team members who can produce quality work right from the beginning, regardless if you’re an entry-level data scientist entering their first job or a seasoned veteran entering a mid-level management position.
In my experience, providing samples of your work whether formally or informally is the right way to present the quality of your work, especially when you’re first starting out. While recruiters may have to take a leap of faith since you have no recommendations behind you, the quality of your work should be able to speak for itself and provide the recommendation of quality that you need.
How to improve this
The best way to prove your track record of quality is through a pile of glowing recommendations from previous instructors, team members, and employers.
If that’s not possible due to you just starting out in the field, you’re going to have to showcase the quality of your work through projects and a portfolio. Your portfolio is how recruiters will be able to track your quality by getting to read about the impact of your projects and see the level of code and documentation that you write.
The data isn’t always good, the business problem isn’t always clear, and the tools are never the same year to year.
Data scientists have to be masters at adapting to all of these occurrences and more. Recruiters are looking for candidates who can roll with the punches and don’t need a pristine environment in which to do their work.
How to improve this
The problem with university data science programs is (as mentioned in point 8) that they often give you simple, squeaky-clean problems to solve where the data isn’t really a mess, the problem is clear, and you have a required set of tools to use that you have been using for the last four years.
This means that to become a more adaptable data scientist, you have to find the worst data set, the most unclear business problem, and use an unfamiliar tool to pull it all together into a coherent analysis. This will be the only way that you can develop your adaptability as a data scientist. It’s not fun, but it will make any problem the company throws at you look like a cakewalk. As mentioned previously, practice will be your best friend in helping you develop this skill.
Adding these projects to your portfolio is a great way to show recruiters that you’re willing to do the hard work and that you can produce results despite adverse conditions. Talking through these projects can also give recruiters a sense of your thought process and how you come around to solve complex problems by adapting to the situation.
There’s a reason why “The Pragmatic Programmer” is such a good book name: it refers to a programmer who can work through problems sensibly and realistically in a manner based on practicality.
Individuals new to data science tend to get wrapped up in the theoretical. This can distract from focusing on one simple thing: getting the project done. At the end of the day, all that matters is that you keep your code simple, make sure your project is adding value to the company, and get the task completed one bite at a time. Nobody cares about your code style if it’s too complex to refactor, nor will anyone care if your project contains a feature that ruins its intended goal. Data science, unlike other tech disciplines, is about getting the job done simply, on time, and while telling one heck of a story that clients and C-level executives will understand and abide by.
Recruiters are looking for data scientists who are practical, sensible, realistic, and determined. Why do an already hard job harder than necessary? Just because you can, doesn’t mean you should. Recruiters will be looking for candidates who can practically work through a problem and produce a conclusion that provides a clear, unaltered picture of what the data is representing. While this is the time when you can show off your practical programming by using one line of code instead of two to get the job done, it shouldn’t be at the expense of the project.
In other words, throw away all the theory that you don’t need to understand to get the job done, and instead focus on the practical aspect — does this mathematical construct work for the question I’m trying to answer? can this code be sent into a production environment? will the stakeholder understand the gravity of the data if I use this visualization?
How to improve this
Pragmatism is well learned by individuals who complete 2-year technical diploma programs or bootcamps because these learning opportunities are often taught by instructors who have spent time in the industry cutting their teeth and truly using their skills. Pragmatism may fall by the wayside for individuals who complete university degrees taught by professors who may have never seen the outside of their faculty buildings.
Pragmatism is something that can be self-taught, and truly boils down to asking yourself this question every time you embark on a new project: what is the bare minimum foundation of information I need to complete this project? Then, you lay out the tools you need from your toolbox: Python and its packages, MySQL, linear regression, and Tableau, for example.
Nothing more and nothing less. Just the bare minimum you need to complete a project.
Yes, you could get into the nitty-gritty of visualization design, or debate the merits of linear regression for this type of analysis until the dogs come home, but at the end of the day, you need to begin working on the project, whether everything is perfect or not. If things need fine-tuning along the way, so be it. But the vital thing here is that you begin. I remember going to school with some individuals who had a hard time beginning projects because their pre-planning wasn’t absolutely perfect. While I believe that 10 minutes of planning can save 10 hours of reworking project aspects, I also understand that, in the fast-paced world of data science, you sometimes just need to start. So, start with what is practical, sensible, and reasonable, and the rest will fall into place.
Learning is difficult, if not impossible without these attributes.
When recruiters are looking for data scientists to join their team, they’re looking for someone who can still learn a new trick here and there. A data scientist’s value to their company is measured by the impact they provide in solving business problems. Therefore, the ability to keep up with new technologies, understand new problems that arise in the sector you’re in, and the ability to learn something new are all skills that require some honesty and humility and that will allow you to provide an impact on your company.
Not only that, but data scientists need to have the honesty and humility required to acknowledge when their analysis isn’t the best it possibly could be or is even wrong. Growth and learning can’t happen without these attributes, which means that recruiters are looking for someone who can keep growing as a data scientist with their company long term.
How to improve this
My favorite way to exercise my honesty and humility is to participate in code reviews and after-project reports. Nothing helps you grow like hearing exactly what you did wrong and how you could have done it better. Code reviews are the perfect way to get input from your colleagues about different techniques to try or alternatives to help your code fit better into the production environment, whereas after-project reports help you see how your work fits into the bigger picture of an entire project. This is especially helpful when you’re working on only a small part of a much bigger project.
Beyond that, the only way to grow as a data scientist is to exercise a continuous feeling of self-improvement toward your skills in the field. New technologies are always cropping up and they’re always teaching something new in MOOCs that could help you become a better data scientist. The key is to make bettering yourself a priority that you incorporate into your daily schedule.
The business problems you’re given to solve in university or a data science bootcamp are squeaky-clean, clear-cut, simple, and almost give you the answer as to how to solve them.
However, once you enter the work world, you’ll realize that the business problems you find there are much harder to translate into analyses. Sometimes the data sucks, or the client has no idea what information they’re really after or a whole variety of wrenches that can get thrown into the mix of your project.
Recruiters will be looking for you to be able to get a clear picture of what the client is looking for, determine if the data you have can be used to solve the problem (and if not, what data you will need from the client), and how you can produce the correct analysis mathematically that will answer your client’s problems. This means being able to work through customer-related problems, optimization problems, predictive problems, and more. Sometimes, clients won’t know what they want and will ask you to solve all of the above-mentioned problems. Your job is to figure out what the real business case is and explain to them why and how your analysis will provide the results they’re looking for to change the trajectory of their company.
Remember: recruiters are looking for you to be attentive and detail-oriented when working with clients, while also having the willingness to tell customers what their real business problem is.
How to improve this
The only way to get better at translating business problems into analyses is to get your hands dirty and work at it. Whether this means solving your own business problems using one of the countless data sets available on the internet, doing pro bono work for a small business in your area, or just rolling up the sleeves when your data science team starts on a new project, the main focus should be on exposing yourself to as many different types of data science problems as possible.
Here’s how to develop the qualities that will help your application stand out
Landing a job in data science can sometimes feel like throwing darts at a board blindfolded — sometimes you’re way off the board in your application and don’t even get an initial interview, sometimes you land close to the bullseye only to not fit the candidate mold perfectly in the technical interview, and sometimes you hit the bullseye and land a job offer.
Sometimes there’s no rhyme or reason for how you did when throwing your dart blindfolded.
This means that it can be equal parts unnerving, frustrating, and crazy when you go to throw your hat in for a job that you can see has hundreds of applicants. You know the type of job I’m talking about — some basic entry-level position for a mid-level company with hundreds of data scientists clamoring for their shot to be discovered. This may be the type of job that you apply for just because it’s there with no real belief that you’ll get it.
However, what if you knew what recruiters were looking for? What if you knew that you could improve your dart-throwing ability by just presenting recruiters with what they’re after? What if you knew that of the hundreds of possible candidates, you and only a handful of others are the real competitors who hold all of the attributes that recruiters are dying for on their teams?
You’d probably go into the application process with a lot more confidence, wouldn’t you?
Over the last several months, I’ve paid attention to recruiter posts on LinkedIn and Reddit that have outlined their struggles in finding ideal data science candidates. Recruiters from varying backgrounds, methodologies, and company sizes have reported the same consistent problems with candidates. From these descriptions, I’ve been able to put together the ideal attributes that data science candidates should have according to the discrepancies noticed during the hiring process, and have coupled these attributes with advice from my own personal experience that you can use to achieve these attributes. Your job search process will no longer involve blindfolded dart-throwing — you just need to focus on developing these eight attributes to help you become the well-rounded data science candidate recruiters are looking for.
It’s estimated that 40% of job candidates lie on their resumes and three out of four recruiters have caught someone lying on their resumes. These are pretty substantial numbers when you consider that hundreds of candidates may be applying for the same data science role at one time. The most common lies found on a resume include education, the time spent with a previous employer, and past salaries. Further lies can be found where candidates have exaggerated impact numbers (such as increasing a company’s data pipeline efficiency by 75%), used inaccurate job titles, lied about their technical abilities, claimed programming (and spoken) language fluency, or even used a fake address.
Therefore, the key here is to always provide recruiters with a recent, updated, and accurate resume that is designed with the job description in mind, represents your current qualifications, and is free from lies or other information that will keep you from getting the job. This may sound easy, but when you’re applying for hundreds of jobs it can sometimes be easy to fall into the trap of sending the same resume to all the companies and hoping one sticks.
Each time I apply for a job, I craft a unique resume keeping the parts that are relevant to the job and rewording sections to make them more applicable to the company. I rarely craft a cover letter unless it’s required to apply, and instead, focus on developing the best one page of information I possibly can. I also focus on a non-graphical resume to ensure that it doesn’t run into parsing issues during the application process. With recruiters only giving a few seconds to your resume, I try to keep it short, sweet, and focused on my ability to produce impact rather than the certificates I’ve amassed or courses I’ve completed.
How to improve this
It’s vital to take the time to produce a tailored resume for each job. This not only shows that you have an understanding of the job requirements but it also shows the recruiter that you have attention to detail and genuinely want the job. Your resume should include all relevant job experience, your past education, certificates you’ve completed, and a list of personal projects that are relevant to the job.
I became a favorite to have on a team for group projects when I was in university studying software development because I was a ringer for presentations who could communicate the value of our project to a wide range of audiences.
I got my first job in tech because I had superior written communication skills that allowed me to explain technical concepts to non-technical clients and potential clients and why they would benefit their company.
Throughout my time working in tech, I was on the front lines of communicating with clients because they felt comfortable talking with someone who could go between them and the technical team to ensure that business problems were being correctly translated into technical cases and vice versa.
In short, I got to where I was in tech by having communication skills.
Data science recruiters are particularly looking for candidates who know what they’re doing inside and out and can answer a variety of questions in interviews that are applicable to daily work. For example, recruiters are looking for candidates who can answer questions in their technical interviews about what they’re doing that go far beyond “Oh, I don’t know, it’s probably in the documentation”.
How to improve this
Unfortunately, communication skills aren’t focused on or taught in traditional data science learning pathways, which means that candidates who succeed in this area have these skills naturally. While communications courses are often taught as part of the program in university or college, one or two courses over 2–4 years generally aren’t enough to make you a competent or comfortable communicator.
Therefore, the only way to develop better communication skills is to practice. My favorite way to practice is to tell my very begrudging dog about everything that I’m doing, from why I’m writing the code the way I am to why I’m using this mathematical construct instead of another. Treating my dog like a recruiter sitting in on a technical interview is not only a great way to practice communicating what I’m working on, but it also gets you in the rhythm of talking while working. I went to school with lots of people who can only code in perfect silence. Even taking breaks to describe what they’re doing would throw them off because they had never practiced.
While some companies may not mind whether or not you can describe what you’re doing while you’re doing it, they’ll certainly want you to be able to give a full rundown at the end, as well as be able to descriptively answer any questions they throw at you beyond, “Oh, I’m sure you can find that in the documentation”.
Because the tech industry is on a sliding scale of requiring and not requiring credentials for a job, you must have the skills to support the job, regardless of whether your credentials are what they should be. This is often the reason why you’ll see hundreds of applications for a single data science job — because the field doesn’t always require credentials, anyone can apply to the job.
This results in many applicants not having the skills to support the job despite applying anyway. Because let’s be honest — how many of us have applied for jobs saying we’re Excel gurus without knowing much beyond =SUM()?
Recruiters seem to be coming across many instances of people not having the right skills for the job (see point 1 above about lying on your resume) which means that while a job may have hundreds of applicants, there may only be 10 or so actual hopefuls.
Luckily, the skills required for the job are always well-outlined in the job description which can give you a good starting point for your studies.
How to improve this
Given that you’re reading this article, I can assume that you’re either a trained data scientist trying to find out what recruiters are looking for, or you’re someone studying data science who’s trying to find out what recruiters are looking for so you can tailor your learning experience. From these assumptions, I can already tell that you probably have the skills for the job that you’re applying to.
The real issue that arises here is with people applying to data science jobs who have no real skills in the area.
Instead of being this person who wastes the recruiter’s time, just read the job description and ensure that you have the skills required. One of my favorite activities is reading job descriptions to see what companies are looking for at a given time. If I don’t have all the skills they’re looking for, I make a list and then begin learning some of them. I do this for all the job ads that I look at so at the end of the learning process, I’ll be well positioned to apply for any data science job that comes across my path.
Not sure where to begin? Try these resources to get started:
I’ve said it before and I’ll say it again: recruiters are only looking for candidates who are capable of producing an impact on the company the moment they join the team. One of the key determinants of this is a candidate’s track record of quality.
There’s not much more to say here beyond the fact that recruiters are looking for team members who can produce quality work right from the beginning, regardless if you’re an entry-level data scientist entering their first job or a seasoned veteran entering a mid-level management position.
In my experience, providing samples of your work whether formally or informally is the right way to present the quality of your work, especially when you’re first starting out. While recruiters may have to take a leap of faith since you have no recommendations behind you, the quality of your work should be able to speak for itself and provide the recommendation of quality that you need.
How to improve this
The best way to prove your track record of quality is through a pile of glowing recommendations from previous instructors, team members, and employers.
If that’s not possible due to you just starting out in the field, you’re going to have to showcase the quality of your work through projects and a portfolio. Your portfolio is how recruiters will be able to track your quality by getting to read about the impact of your projects and see the level of code and documentation that you write.
The data isn’t always good, the business problem isn’t always clear, and the tools are never the same year to year.
Data scientists have to be masters at adapting to all of these occurrences and more. Recruiters are looking for candidates who can roll with the punches and don’t need a pristine environment in which to do their work.
How to improve this
The problem with university data science programs is (as mentioned in point 8) that they often give you simple, squeaky-clean problems to solve where the data isn’t really a mess, the problem is clear, and you have a required set of tools to use that you have been using for the last four years.
This means that to become a more adaptable data scientist, you have to find the worst data set, the most unclear business problem, and use an unfamiliar tool to pull it all together into a coherent analysis. This will be the only way that you can develop your adaptability as a data scientist. It’s not fun, but it will make any problem the company throws at you look like a cakewalk. As mentioned previously, practice will be your best friend in helping you develop this skill.
Adding these projects to your portfolio is a great way to show recruiters that you’re willing to do the hard work and that you can produce results despite adverse conditions. Talking through these projects can also give recruiters a sense of your thought process and how you come around to solve complex problems by adapting to the situation.
There’s a reason why “The Pragmatic Programmer” is such a good book name: it refers to a programmer who can work through problems sensibly and realistically in a manner based on practicality.
Individuals new to data science tend to get wrapped up in the theoretical. This can distract from focusing on one simple thing: getting the project done. At the end of the day, all that matters is that you keep your code simple, make sure your project is adding value to the company, and get the task completed one bite at a time. Nobody cares about your code style if it’s too complex to refactor, nor will anyone care if your project contains a feature that ruins its intended goal. Data science, unlike other tech disciplines, is about getting the job done simply, on time, and while telling one heck of a story that clients and C-level executives will understand and abide by.
Recruiters are looking for data scientists who are practical, sensible, realistic, and determined. Why do an already hard job harder than necessary? Just because you can, doesn’t mean you should. Recruiters will be looking for candidates who can practically work through a problem and produce a conclusion that provides a clear, unaltered picture of what the data is representing. While this is the time when you can show off your practical programming by using one line of code instead of two to get the job done, it shouldn’t be at the expense of the project.
In other words, throw away all the theory that you don’t need to understand to get the job done, and instead focus on the practical aspect — does this mathematical construct work for the question I’m trying to answer? can this code be sent into a production environment? will the stakeholder understand the gravity of the data if I use this visualization?
How to improve this
Pragmatism is well learned by individuals who complete 2-year technical diploma programs or bootcamps because these learning opportunities are often taught by instructors who have spent time in the industry cutting their teeth and truly using their skills. Pragmatism may fall by the wayside for individuals who complete university degrees taught by professors who may have never seen the outside of their faculty buildings.
Pragmatism is something that can be self-taught, and truly boils down to asking yourself this question every time you embark on a new project: what is the bare minimum foundation of information I need to complete this project? Then, you lay out the tools you need from your toolbox: Python and its packages, MySQL, linear regression, and Tableau, for example.
Nothing more and nothing less. Just the bare minimum you need to complete a project.
Yes, you could get into the nitty-gritty of visualization design, or debate the merits of linear regression for this type of analysis until the dogs come home, but at the end of the day, you need to begin working on the project, whether everything is perfect or not. If things need fine-tuning along the way, so be it. But the vital thing here is that you begin. I remember going to school with some individuals who had a hard time beginning projects because their pre-planning wasn’t absolutely perfect. While I believe that 10 minutes of planning can save 10 hours of reworking project aspects, I also understand that, in the fast-paced world of data science, you sometimes just need to start. So, start with what is practical, sensible, and reasonable, and the rest will fall into place.
Learning is difficult, if not impossible without these attributes.
When recruiters are looking for data scientists to join their team, they’re looking for someone who can still learn a new trick here and there. A data scientist’s value to their company is measured by the impact they provide in solving business problems. Therefore, the ability to keep up with new technologies, understand new problems that arise in the sector you’re in, and the ability to learn something new are all skills that require some honesty and humility and that will allow you to provide an impact on your company.
Not only that, but data scientists need to have the honesty and humility required to acknowledge when their analysis isn’t the best it possibly could be or is even wrong. Growth and learning can’t happen without these attributes, which means that recruiters are looking for someone who can keep growing as a data scientist with their company long term.
How to improve this
My favorite way to exercise my honesty and humility is to participate in code reviews and after-project reports. Nothing helps you grow like hearing exactly what you did wrong and how you could have done it better. Code reviews are the perfect way to get input from your colleagues about different techniques to try or alternatives to help your code fit better into the production environment, whereas after-project reports help you see how your work fits into the bigger picture of an entire project. This is especially helpful when you’re working on only a small part of a much bigger project.
Beyond that, the only way to grow as a data scientist is to exercise a continuous feeling of self-improvement toward your skills in the field. New technologies are always cropping up and they’re always teaching something new in MOOCs that could help you become a better data scientist. The key is to make bettering yourself a priority that you incorporate into your daily schedule.
The business problems you’re given to solve in university or a data science bootcamp are squeaky-clean, clear-cut, simple, and almost give you the answer as to how to solve them.
However, once you enter the work world, you’ll realize that the business problems you find there are much harder to translate into analyses. Sometimes the data sucks, or the client has no idea what information they’re really after or a whole variety of wrenches that can get thrown into the mix of your project.
Recruiters will be looking for you to be able to get a clear picture of what the client is looking for, determine if the data you have can be used to solve the problem (and if not, what data you will need from the client), and how you can produce the correct analysis mathematically that will answer your client’s problems. This means being able to work through customer-related problems, optimization problems, predictive problems, and more. Sometimes, clients won’t know what they want and will ask you to solve all of the above-mentioned problems. Your job is to figure out what the real business case is and explain to them why and how your analysis will provide the results they’re looking for to change the trajectory of their company.
Remember: recruiters are looking for you to be attentive and detail-oriented when working with clients, while also having the willingness to tell customers what their real business problem is.
How to improve this
The only way to get better at translating business problems into analyses is to get your hands dirty and work at it. Whether this means solving your own business problems using one of the countless data sets available on the internet, doing pro bono work for a small business in your area, or just rolling up the sleeves when your data science team starts on a new project, the main focus should be on exposing yourself to as many different types of data science problems as possible.