Should a data scientist resume list Kaggle or side projects?

Include them when they show something the job history does not, such as a domain you want to move into or a method you have not used at work. Keep them brief and lead with what you learned or decided.

Do I need a publications section?

Only if it is relevant to the role. For research-heavy positions it carries weight; for product-focused roles a short line is enough and results from real projects matter more.

Are the numbers in this example real?

No, the person, employers and figures are illustrative. Use figures you can explain, including how the metric was measured.

How much math should the resume show?

Enough to show you know what the method assumes. Mentioning held-out data, power calculations or a baseline signals that more efficiently than listing techniques.

Data Scientist Resume

Data Scientist Resume Example

A data scientist resume example that connects models to decisions, with a note on why each bullet works and how to avoid the common mistakes.

Full example resume

Line-by-line notes

Before and after rewrites

Free, no sign-up

Analyze My Resume Talk To Us

Illustrative example. The person, employers and every figure below are fictional. The structure is what to copy: a result next to each claim. Use your own numbers, and only ones you could explain if asked how you measured them.

Priya Nair

Data Scientist

Boston, MA | priya.nair@example.com | linkedin.com/in/priya-nair-example

Summary

Data scientist with five years turning messy product and customer data into decisions, using Python, SQL and applied statistics. Strongest at framing the question before modelling it, and at explaining results to people who do not read code.

Experience

Data Scientist - Consumer subscription company, Boston2021 - Present

Built a gradient-boosted churn model in Python and Scikit-learn, reaching 0.81 AUC on a held-out quarter, and worked with the retention team to target outreach, which reduced monthly churn from 4.6% to 3.9%.
Designed and analysed 14 A/B tests on onboarding, using pre-registered success metrics and power calculations; three shipped changes raised week-one activation by a combined 11%.
Replaced a manual weekly reporting process with a scheduled pipeline in SQL and Pandas, returning about six analyst-hours a week and removing recurring copy errors.
Presented findings to product and finance leadership each quarter, including a recommendation not to launch a feature after a 3-week experiment showed no effect, which saved an estimated two engineering sprints.

Data Analyst - Healthcare services firm, Boston2018 - 2021

Built Tableau dashboards for clinic operations used by 30+ managers, replacing weekly spreadsheet exports.
Used regression analysis to identify the scheduling factors most associated with no-shows, informing a reminder change that reduced no-shows by 7%.
Cleaned and reconciled four source systems into a single reporting table, documenting every one of its 60 assumptions for later audit.

Skills

Core: Python, SQL, Machine Learning, Statistics, Pandas, NumPy, Scikit-learn, Data visualization

Tools: Tableau, Airflow, dbt, A/B testing, Spark

Education

M.S. Statistics, State University, 2018

Weak bullets, rewritten

The most common data scientist bullets and what a stronger version looks like. Each rewrite adds the thing the weak version leaves out.

Before

Built machine learning models to predict customer behaviour.

After

Built a churn model reaching 0.79 AUC on held-out data and used it to prioritise 5,000 retention calls a month, reducing cancellations by 8%.

Why it works: The rewrite pairs a rigour signal (held-out data) with a business effect, so a reader knows the model was both sound and used.

Before

Analysed data and presented insights to stakeholders.

After

Showed the finance team that a planned discount would cost $180K a quarter with no measurable lift, and it was withdrawn before launch.

Why it works: 'Presented insights' describes a meeting. The rewrite describes a decision that changed because of the analysis.

Before

Proficient in Python, SQL and statistics.

After

Wrote the SQL and Python that reconciled four billing systems, then used a bootstrap interval to show the discrepancy was real rather than noise.

Why it works: Skills are claims; the rewrite demonstrates two of them inside one piece of work.

Once your own version is written, run it through the ATS parser to confirm your titles and dates come through in order, then compare it with the posting using resume job match.

The summary

It puts the differentiator, framing the question before modelling it and explaining results to non-technical people, ahead of the tools. Tools are what most candidates list; judgement is what few do.

The churn bullet

The model metric (0.81 AUC on a held-out quarter) proves rigour, and the second half (churn from 4.6% to 3.9%) proves the model changed something. Either half alone would be weaker.

The A/B testing bullet

Pre-registered metrics and power calculations are the details a statistician looks for. They show discipline about false positives, which is a core professional risk in the role.

The "do not launch" bullet

A recommendation against shipping is unusually strong evidence. It shows integrity and the willingness to deliver a null result, which hiring managers trust more than a string of wins.

The analyst role

It shows progression from reporting into analysis into modelling, with the reason each step happened visible in the bullets.

Related Resume Pages

Use these pages to keep moving through the same topic cluster instead of bouncing back into generic advice.

All Resume Examples Machine Learning Engineer Resume Example Site Reliability Engineer Resume Example Security Analyst Resume Example ATS Parser Online Resume Job Match

A data scientist resume is judged on one question that most of them never answer: what changed because of the model? Accuracy figures on their own read as coursework. The example below ties each piece of analysis to a decision someone made, which is what a hiring manager is actually buying.

Recommended Workflow

Step 1

Match the problem type

A growth role wants experiments and causal thinking; an operations role wants forecasting and optimisation. Reorder bullets so the closest problem type leads.

Step 2

Use the posting's tooling where you have it

If the role names Spark, dbt or PyTorch and you have used them, put them in the bullet that used them, not only in the list.

Step 3

Compare against the posting

Paste the job description into the resume job match tool and see which required terms your resume never states, then add those you can evidence.

Common Mistakes This Page Can Help You Catch

Reporting accuracy without a decision

A model at 92% accuracy tells a reader nothing about whether it was used. Always attach what someone did differently because of it.

Listing every library ever imported

A skills block of thirty names reads as a course syllabus. Keep the ones you use weekly, and let projects and results show the rest.

Hiding the messy parts

Data cleaning, reconciliation and documentation are most of the job. Mentioning them honestly signals real experience; omitting them reads like an academic profile.

Frequently Asked Questions

Next Step

Turn Resume Advice Into A Better Application

Use the free analyzer to get your ATS score, then move into job match, rewrite, and cover letter workflows when you are ready to tailor applications faster.

Analyze Your Resume