How to Write A Data Science Resume
The best data science resumes have a lot in common. Here is a detailed guide to building an effective resume.
Vin Vashishta | Originally Published: December 22nd, 2020
6 years. That is how long I have been hiring for data science and machine learning roles. I built the models that run a resume screening and candidate selection application. I have researched hiring and how people make hiring decisions.
What have I learned? Most resumes are badly written. The most common problems are:
Hiring has a lifecycle just like model development. Your resume needs to be built to stand out at each phase. What are they?
Automated Resume Screening: Once you submit your resume, it is parsed by an application. Your resume must be formatted to score well and advance to the next phase.
HR and Recruiter Screening: The resume selection application creates a shortlist of candidates for someone in HR or a Recruiter to manually review. They are looking at the role through a semi-technical, sometimes partial understanding of the qualifications. Your resume needs to target their decision-making process.
Hiring Manager: The hiring manager is looking for indicators that you can be successful in the job. They are bombarded by keyword heavy resumes. Most candidates do not get an interview because their resumes focus on skills not capabilities.
A well-built resume results in recruiters reaching out to you rather than you applying for hundreds of jobs to only get a few responses. Automated resume screeners are looking through resumes, LinkedIn profiles, Github repos, and other external sources daily. Even without submitting your resume, you should be getting regular emails and phone calls for jobs.
If that is not happening, it is time to rebuild your resume.
Resume screeners do 3 things:
Resume screeners are a lot like search engines and there is a resume version of optimization for discovery. Formatting is especially important. Parsers range from a massive RegEx to NLU based systems. Your resume needs to be written so both can read it.
If a parser cannot find contact information, your resume will be dropped from the search results. If the education section is poorly labeled or formatted, a dumb parser will not pick up your degree and you will be dropped from the search results.
NLU based parsers are looking for semantics. Complete sentences are important. A well written project description ranks higher than a skills list. Writing for smart parsers is like writing for recruiters and HR screeners.
The first real person who will read your resume is probably a recruiter or someone in HR. Make their job easy because in many companies, they are the gatekeepers between you and the hiring manager. The first major section in your resume is a capabilities section. This is where you map the connection between your skills and the job requirements. It is a list using plain language.
What these first line screeners are looking for is ability. The feedback they most often get from hiring managers is, “I am looking for hands on experience with, not just knowledge of.” The first section in your resume needs to focus on differentiating yourself from those with textbook knowledge.
If you are a recent graduate or transitioning into the field, this section is the most important in your resume. Regardless of experience, everyone needs to overcome 2 hurdles to get past the HR or recruiter filter.
They are reading to exclude your resume. The reward of spending more than a minute on each resume is low so they speed through, looking for what typically knocks people out of consideration. You need to have each key point covered: education, experience, and applied knowledge.
They have semi-technical knowledge. They are familiar with high level concepts: deep learning, machine vision, python, tensorflow, and statistics. They may not be familiar with decision trees, GANs, transformers, GPU optimization, or project methodologies.
The first section in your resume needs to contain a mixture of language. Introduce each main point with a simplistic explanation using high level terms. Make sure that is the first sentence of each section. Put details in the following sentences. Use this pattern anywhere you provide detail about your experience and project work but especially in this first section.
Make sure someone scanning your resume cannot miss your key capabilities. Why have education on your resume in a capabilities and a dedicated section? It is less likely someone glancing at your resume will scan past it.
Your personal contact information should be at the top of your resume: Name, location, phone number, email address, personal website link, and GitHub repo link. The last two are critical. Where can someone go to see your project work? Where can they go to learn more about you?
You can only put so much into a resume. A GitHub link or personal website allows a hiring manager to learn more. A few blog posts, a more detailed bio, and writeups of your most recent or relevant projects are all that you need here. Some hiring managers hate resumes. A more detailed alternative will help you get an interview.
The next section is the above-mentioned capabilities section. For a hiring manager, the supporting sentences are important. Use active language and specifics. Focus on real world challenges and solutions. Put in a line or 2 about the team(s) you worked with and how effectively the team collaborated to produce a tangible result.
The order of the remaining sections should play to your strengths. Your choices are project work, experience, and education. You will quickly lose a hiring manager’s attention so this may be the only other section they read. Order is important.
For experience, give each job its own paragraph. Header each job with your job title, company name, and date range. Location is optional. The body of each job should list projects worked on. With each project include skills used, work products created, inputs used to build those work products, and outcomes achieved.
If you led a team or organization, the outcomes are team and personal outcomes. What did the team accomplish under your leadership and what did you accomplish?
In some cases, your job title and responsibilities do not match up. Here is how to simply indicate that. Job Title: Data Scientist, Functional Role: Machine Learning Engineer.
Project work follows the same format and content as work experience. Change the header from company or role to the project title. Emphasize outcomes; what did the project result in? Connect technologies used to applications and work products.
Education is straight forward: where you went to school, degree attained, and honors or awards. Create a section for publications, awards, and speaking appearances if any of those apply to you.
If you are applying directly to a job, customize your resume to speak to the hiring manager. Show that you have:
This last one is especially important if you have mostly independent project work on your resume. Show that work resulted in something significant; beyond you becoming familiar with that kind of project.
Think of this as empathizing with your hiring manager. Tell them you know what they are working on and trying to accomplish. Give them a sense that you have been in their shoes. You want 1 or 2 lines to get a hiring manager to think to themselves, “You are going to need that here.”
That is a lot of work. You want to target and spend the time on a small number of jobs. At the same time, you want to cast a wide net. A well written resume does passive job hunting for you. Applications are scrubbing websites for resumes and candidate information.
LinkedIn, Monster, Indeed, and GitHub are the most scraped sites. Some candidate discovery happens on social media, Stack, and Kaggle. However, those are less common. Your resume and profile will get recruiters to reach out directly. If that is not happening, it is time to rewrite your resume until they do. Always doubt your resume before you doubt your capabilities.
Update your resume monthly. Continuously review and improve it. Every time you change your resume, the last updated field pushes it to the top of the pile again. The older the resume, the lower it may score. Some resume discovery tools use resume age as an indication of how accurate the resume is or how likely you are to be open to a new role.
Happy job hunting.