Data scientist and machine learning engineer CV example

Begin with a model inventory

List every model you trained, tuned or put into service in the last few years, including the ones that were shelved. For each, fill in the same six fields. Hiring panels for these roles ask about exactly these facts, and a line missing one of them is the line they will question.

1. Task: what the model predicted or ranked, in business words (late payment, product search, demand per store).
2. Data: rows, features or tokens, and how fresh the training set was.
3. Metric: the one the team agreed on, with its value on held-out data.
4. Baseline: what you compared against, such as the old rule, the previous model or a simple heuristic.
5. Status: notebook only, offline evaluation, A/B test, or serving live traffic.
6. Your share: designed, trained, deployed, monitored, or reviewed.

A model with no baseline cannot be judged; a model with no status leaves the reader guessing whether it ever left your laptop. Fix both before you write a single bullet.

Data scientist or ML engineer: which CV are you writing

The UK government’s Digital and Data capability framework separates the two. Its data scientist role spans maths, statistics, software engineering and communication, and expects people to use different performance and accuracy metrics for model validation. Its machine learning engineer role develops, assures and maintains models in products, deploys them into production against performance requirements, and checks that live models keep working safely.

Many roles sit in between. Choose the column that matches the posting’s first three requirements and order your lines accordingly; the facts stay the same.

What employers look for first

The US Bureau of Labor Statistics describes data scientists as deciding which data are useful, collecting and analysing it, building, validating and testing models, presenting results and advising decision makers. It says a bachelor’s degree in mathematics, statistics or computer science is typically the minimum and that some employers want a master’s or doctorate. In O*NET’s survey for the occupation, 48% of respondents said a bachelor’s is required and 44% a master’s.

• Education with quantitative depth: Prospects lists computer science, data science, mathematics, physics and statistics, and MSc or PhD routes.
• A route without the classic degree: the National Careers Service lists a Data Scientist Level 6 degree apprenticeship and master’s conversion courses.
• The languages: the National Careers Service names R, SQL and Python; O*NET’s software list includes TensorFlow, PyTorch and Amazon SageMaker.
• Validation: O*NET’s task list includes testing, validating and reformulating models so they predict accurately.

Is a PhD required

Not by the official profiles we read, which present the doctorate as one route among several. For research scientist posts a PhD is often the norm; the posting will say so in its first requirement.

Writing a model line with its metric

Use the metric the problem was judged on

The scikit-learn documentation gives the plain rule: if a scoring function is set, as in a Kaggle competition or a business context, use that one; if you are free to choose, start from the ultimate goal of the prediction. Your CV should report the agreed metric, not the most flattering one.

Accuracy on imbalanced classes

A fraud model that is 99% accurate on data with 1% fraud may have learned nothing. scikit-learn notes that balanced accuracy avoids inflated estimates on imbalanced datasets. Report recall at a stated precision, PR AUC or balanced accuracy, and the class ratio.

Offline score, then business result

Example: data scientist CV section

Example only. Deniz Kaplan is invented, and so are her employers, metrics and paper. Each line carries task, metric, baseline and status, and the publication entry gives enough detail for a reader to find it.

Senior Data Scientist, a consumer lending company, 2023 to present

• Rebuilt the early-default model on 2.1 million loans with LightGBM; recall at 80% precision rose from 0.38 to 0.51 against the logistic baseline, and credit policy adopted it for new applicants.
• Added monthly drift and fairness checks by age band and region, reported in a model card reviewed by risk and legal before each release.
• Mentored two data scientists through their first production models.
• Tools: Python, LightGBM, scikit-learn, SQL (Snowflake), MLflow, Airflow.

Data Scientist, an online travel agency, 2020 to 2023

• Trained the hotel re-ranking model; a three-week A/B test raised booking conversion from 2.6% to 2.8% and it was rolled out to all markets.
• Replaced a weekly hand-built demand spreadsheet with a forecasting job for 35 cities (MAPE 12% vs 19%).

Publications and competitions

• Kaplan, D. and Öztürk, S. (2022). “Calibrated default prediction under distribution shift.” Workshop paper, with link to the published PDF.
• Kaggle: solo entry, top 4% of 1,900 teams in a tabular forecasting competition (2021), profile link included.
• MSc thesis, Statistics, 2020: Survival models for loan prepayment.

Papers, theses and Kaggle on a CV

Prospects says competitions such as Kaggle and Topcoder can help you gain visibility and experience. They help a CV only when the entry is specific and checkable. Write your own rank, the field size, whether you competed alone or in a team, the year, and a link. A team medal you did not work on, or a vague "Kaggle Expert" with no profile, invites a question you cannot answer.

• Papers: authors in order, year, title, venue and a link. Mark preprints as preprints.
• Theses: degree, subject, year and title; add a link if the full text is public.
• Model documentation: if you wrote model cards, say so. The model cards paper by Mitchell and colleagues asks for evaluation across relevant groups and the intended use, which is exactly the evidence a reviewer wants.
• Keep personal projects that taught you a method only until you have two production lines to replace them.

Skills table for data science and ML engineering

Licences and cloud certificates

No licence to practise appears in the BLS, O*NET, National Careers Service or Prospects profiles. Prospects mentions accreditation as a Data Science Professional from the Alliance for Data Science Professionals after at least two years in a relevant role. Cloud certificates help ML engineers when a posting names that cloud.

Proof: sources and the date we read them

All pages were read on 13 October 2026. Links are under Sources.

Where this page stops

• Reporting, dashboards and SQL-first roles: the data analyst guide.
• Infrastructure as code and cloud operations as the main job: the DevOps guide.
• Graduates with a computer engineering degree writing a first CV: the computer engineer guide.

Applying abroad and next steps

Residence rules sit in the country guides: degree recognition and the salary threshold in the EU Blue Card guide, IND salary bands in the Netherlands highly skilled migrant guide, and sponsorship in the UK guide for international applicants.

1. Complete the six-field inventory and drop models with no baseline you can name.
2. Decide which column of the scientist versus engineer table the posting wants.
3. Write each line as task, metric against baseline, status and use.
4. Add papers, thesis and competitions with links, or leave them out.

CitedCV can write these lines from the facts you enter, in English, Turkish or German, and checks numbers against what you gave it, so a PR AUC on the CV is the one you typed, not a rounder guess.

Questions people ask

Sources

1. U.S. Bureau of Labor Statistics, Occupational Outlook Handbook: Data Scientistswww.bls.gov/ooh/math/data-scientists.htm
2. O*NET OnLine, 15-2051.00 Data Scientistswww.onetonline.org/link/summary/15-2051.00
3. National Careers Service, Data scientist job profilenationalcareers.service.gov.uk/job-profiles/data-scientist
4. Prospects, Data scientist job profilewww.prospects.ac.uk/job-profiles/data-scientist
5. UK DDaT capability framework, Data scientistunderstand-digital-data-roles-skills.service.gov.uk/role/data-scientist
6. UK DDaT capability framework, Machine learning engineerunderstand-digital-data-roles-skills.service.gov.uk/role/machine-learning-engineer
7. scikit-learn, Metrics and scoring: quantifying the quality of predictionsscikit-learn.org/stable/modules/model_evaluation.html
8. Mitchell et al., Model Cards for Model Reporting (arXiv 1810.03993)arxiv.org/abs/1810.03993
9. AWS Certified Machine Learning Engineer, Associateaws.amazon.com/certification/certified-machine-learning-engineer-associate/
10. Google Cloud, Professional Machine Learning Engineercloud.google.com/learn/certification/machine-learning-engineer