Data
Data Scientist CV: what employers ask for
Data scientist postings vary widely, from experimentation to machine learning in production. Screeners want to see which kind you are, what models or analyses you built, how you judged them and whether they made it into real use.
From real postings
We read 86 Data Scientist job postings from 42 companies on 6 November 2026. The most named asks were SQL (77), Python (76) and Statistics (64). 55 postings stated years of experience; the middle value was 5.
One-time analysis • $2.99 • No subscription
What a strong Data Scientist CV looks like
Start here if you are editing a CV today. The evidence for each point follows below.
Section order, top to bottom
- HeaderName, email, phone, city, and a link to work you can discuss, such as a repository or paper.
- SummaryTwo or three lines: your title, years of experience, your focus (experiments, modelling, forecasting), and your tools.
- ExperienceMost recent first. Bullets about models or analyses that people used, with the baseline you beat and who acted on it.
- SkillsGrouped: SQL, Python or R, statistics and experimentation, machine learning libraries.
- Publications or projectsOnly those that match the posting. One line each, with your part.
- EducationDegree and subject. Many postings in our sample mentioned a degree.
The summary
Say what kind of data science you do. Statistics and A/B testing came up more often than machine learning in our sample, so do not hide experiment work behind model names.
Data scientist with four years of experience in experimentation and forecasting. I design A/B tests with product teams and build models in Python and SQL that operations teams plan with.
Skills: show them, do not just list them
SQL and Python were named in most postings we read. Statistics came next, then experimentation, then machine learning. Show each through work that someone used, with the baseline you compared against.
Weak and strong bullets
Illustrative rewrites, not real candidates. Use your own work, and only numbers you can back up.
Weak: Built machine learning models.
Stronger: Built a gradient-boosted churn model that beat the rule-based baseline by 12 points of recall; the retention team now uses its weekly list.
Why it works: A model compared with a baseline, and a team that uses it.
Weak: Ran experiments.
Stronger: Designed the experiment for a new recommendation widget, choosing guardrail metrics that caught a drop in repeat purchases before full rollout.
Why it works: Experiment design with a guardrail that changed a decision.
Weak: Presented insights to stakeholders.
Stronger: Presented demand-forecast results to the operations leads, who used them to set staffing for the holiday peak.
Why it works: Shows communication through an outcome, not a soft-skill claim.
What Data Scientist job postings ask for
Counted from 86 real postings, not guessed. When you paste a posting into Resymetry, it checks each of these asks against your CV and marks it as shown, only listed, or missing.
| Skill or ask | Postings naming it | Only as nice to have |
|---|---|---|
| SQL | 77 of 86 | 1 |
| Python | 76 of 86 | 1 |
| Statistics | 64 of 86 | – |
| A/B testing / experimentation | 49 of 86 | 2 |
| Machine learning | 47 of 86 | 1 |
| Stakeholder management | 45 of 86 | 2 |
| Metrics / KPIs | 42 of 86 | 1 |
| Cross-functional work | 39 of 86 | – |
| Roadmap | 28 of 86 | – |
| R | 27 of 86 | – |
Show 8 more
| Skill or ask | Postings naming it | Only as nice to have |
|---|---|---|
| LLMs / generative AI | 21 of 86 | 1 |
| Monitoring / observability | 17 of 86 | – |
| Spark | 17 of 86 | 2 |
| Data modelling | 15 of 86 | 3 |
| Prioritisation | 15 of 86 | – |
| Prototyping | 14 of 86 | – |
| Looker | 13 of 86 | 1 |
| dbt | 12 of 86 | 3 |
How we counted
On 6 November 2026 we read 86 live Data Scientist postings from 42 companies, through their public job boards (Greenhouse, Lever and Ashby). A posting counted when its title matched the role. We took at most three per company, so no single employer sets the result.
A skill counts once per posting, when it is named outside the benefits and company boilerplate. Matching is by keyword, so read the numbers as “named in”, not “strictly required”. “Only as nice to have” means it appeared only under a heading such as “Nice to have” or “Bonus”. 34 of the 86 postings had such a heading, so that column is a floor.
By location on the posting: 51 US, 9 UK, 3 India, 23 elsewhere or remote. Most employers in the sample are technology companies, which skews the result towards their stacks.
Companies sampled: Airbnb, Asana, Bloomreach, Brex, Carvana, Chime, Cloudflare, Cohere, Coinbase, Coursera, Databricks, Discord, Duolingo, ElevenLabs, Epic Games, Figma, Fivetran, Flexport, Gemini, GitLab, Gusto, Intercom, Lyft, Meesho, Mixpanel, MongoDB, Monzo, Mozilla, Netlify, OpenAI, Oscar Health, Perplexity, Pinterest, Ramp, Reddit, Robinhood, Roblox, Spotify, Stripe, Toast, Wealthfront, Yext.
The script that produced these numbers is scripts/role-postings.mjs in the Resymetry repository. It stores counts only, never the text of a posting.
What employers look for in a Data Scientist CV
Screeners check these first. Each one should be visible in your job entries, not only in a skills list.
- Python with the usual libraries, and SQL for getting your own data.
- Models or statistical analyses that reached real use, with how they were evaluated.
- Experimentation skills: designing tests, choosing metrics and reading results correctly.
- Communicating findings to people who are not data scientists.
- For ML-heavy roles, experience getting models into production with engineers.
4 common Data Scientist CV mistakes and how to fix them
Listing algorithms instead of problems.
Fix: Start with the business problem, then the approach and the result.
Reporting accuracy with no baseline.
Fix: Compare against what existed before, and give the business effect where you know it.
Course and competition projects outweigh work experience.
Fix: Lead with work. Keep only projects that show something your jobs do not.
Not saying whether a model was ever used.
Fix: State whether it went to production, informed a decision or stayed a prototype.
Starting from scratch, or the layout is the problem rather than the content? There are free ATS CV templates to start from.
Know your chances for Data Scientist roles
Your score and your biggest problems are free. The full report is a one-time $2.99.
Data Scientist CV questions
- Should a data scientist CV include Kaggle projects?
- Only if they show a skill your work does not, or if you have little work experience. Describe the problem and result, not just the ranking.
- How do I show the impact of a model?
- Compare it with the previous approach, say who used it and describe the effect you actually observed.
- Do I need deep learning on my data scientist CV?
- Only if the posting asks for it. Many data scientist roles focus on statistics, experiments and classical models.