A complete, ready-to-render data scientist resume in YAML. Copy it, swap in your numbers, generate an ATS-friendly PDF.

A strong data scientist resume does three things: it quantifies model and experiment impact in business terms, it organizes skills so a recruiter can match them to the job description in seconds, and it formats cleanly for both ATS parsers and human readers.

Here's a complete RenderCV input file that does all three. It's a real, working YAML file. Copy it, replace the content with your own, and render.

Complete YAML example

cv:
  name: Priya Nair
  location: Boston, MA
  email: priya.nair@email.com
  phone: "+16175550199"
  website: https://priyanair.io
  social_networks:
    - network: LinkedIn
      username: priyanair-ds
    - network: GitHub
      username: pnair-ml

  sections:
    summary:
      - Data scientist with 5 years of experience building
        production machine learning systems and running
        large-scale experiments. Focused on measurable revenue
        impact, rigorous A/B testing, and shipping models that
        survive contact with real traffic.

    experience:
      - company: DoorDash
        position: Senior Data Scientist
        location: Boston, MA
        start_date: 2022-08
        end_date: present
        highlights:
          - Built a gradient-boosted demand forecasting model
            (XGBoost) that cut delivery time prediction error by
            22%, lifting on-time delivery rate from 88% to 94%.
          - Designed and ran 40+ A/B tests on the dispatch
            algorithm, shipping changes that increased order
            throughput by $14M in annualized revenue.
          - Reduced model serving latency from 180ms to 45ms by
            migrating feature computation to a real-time feature
            store, enabling per-request personalization.
      - company: Wayfair
        position: Data Scientist
        location: Boston, MA
        start_date: 2020-06
        end_date: 2022-07
        highlights:
          - Developed a churn prediction model (logistic
            regression with calibrated probabilities) achieving
            0.84 AUC, driving a retention campaign that recovered
            $3.2M in at-risk revenue.
          - Built an uplift model to target promotional discounts,
            increasing incremental conversion by 11% while reducing
            discount spend by 18%.
          - Standardized the team's experimentation workflow with a
            shared Python library for power analysis and sequential
            testing, cutting test setup time from 2 days to 2 hours.
      - company: Liberty Mutual
        position: Data Analyst
        location: Boston, MA
        start_date: 2019-01
        end_date: 2020-05
        highlights:
          - Built automated SQL pipelines and dashboards for claims
            analytics, replacing manual reporting that took 8 hours
            per week.
          - Modeled claim severity with regularized regression,
            improving reserve accuracy by 9% over the prior
            heuristic.

    projects:
      - name: ab-test-toolkit
        date: 2023-09
        highlights:
          - Python library for A/B test analysis with frequentist
            and Bayesian methods, variance reduction (CUPED), and
            automatic power calculations. 600+ GitHub stars.
          - "[github.com/pnair-ml/ab-test-toolkit](https://github.com/pnair-ml/ab-test-toolkit)"
      - name: churn-baseline
        date: 2022-11
        highlights:
          - End-to-end churn modeling template covering feature
            engineering, calibration, and SHAP-based explanations,
            packaged as a reproducible pipeline.
          - "[github.com/pnair-ml/churn-baseline](https://github.com/pnair-ml/churn-baseline)"

    education:
      - institution: Carnegie Mellon University
        area: Statistics & Machine Learning
        degree: MS
        start_date: 2017-09
        end_date: 2018-12
        highlights:
          - "Relevant coursework: Statistical Machine Learning,
            Experimental Design, Bayesian Inference, Convex
            Optimization"
      - institution: University of Michigan
        area: Mathematics
        degree: BS
        start_date: 2013-09
        end_date: 2017-05
        highlights:
          - "GPA: 3.9/4.0"

    skills:
      - label: Languages
        details: Python, SQL, R
      - label: Machine Learning
        details: scikit-learn, XGBoost, PyTorch, statsmodels,
          time-series forecasting, gradient boosting
      - label: Experimentation
        details: A/B testing, causal inference, uplift modeling,
          power analysis, CUPED, sequential testing
      - label: Data & Infrastructure
        details: PostgreSQL, Snowflake, Spark, Airflow, dbt, MLflow
      - label: Visualization
        details: matplotlib, seaborn, Tableau, Streamlit

design:
  theme: engineeringresumes

This validates against RenderCV's JSON Schema. Every date is parsed and formatted consistently, every section follows a defined structure, and the output is a single-column PDF that ATS systems parse without choking. For more on why that matters, see our guide on building an ATS-friendly resume.

bold_keywords: surface your stack automatically

When you tailor the resume to a specific role, bold_keywords lets you emphasize the relevant tools and methods without editing individual bullets:

settings:
  bold_keywords:
    - Python
    - SQL
    - XGBoost
    - A/B test
    - A/B tests
    - causal inference
    - forecasting
    - feature store
    - PyTorch

Every occurrence of these terms gets bolded in the PDF. A recruiter scanning the page sees your core methods immediately, and ATS parsers still read clean text underneath.

The practical benefit: an experimentation-heavy role might bold A/B test, causal inference, and uplift modeling. A modeling role shifts to XGBoost, PyTorch, and forecasting. You change one list and regenerate. The bullets stay exactly the same.

What makes each section work

Experience

Every bullet follows the same shape: action, technical method, measurable result. "Built a gradient-boosted demand forecasting model (XGBoost) that cut delivery time prediction error by 22%, lifting on-time delivery rate from 88% to 94%." That tells the reader what you built, how, and what it moved. One sentence.

For data scientists, the measurable result is where most resumes fall apart. Two rules make it credible:

Quantify the model AND the business. "0.84 AUC" alone is a lab metric. "0.84 AUC, driving a retention campaign that recovered $3.2M in at-risk revenue" connects the model to a dollar figure a hiring manager cares about. Pair the offline metric with the online outcome whenever you can.
Quantify experiments honestly. "Ran 40+ A/B tests that increased order throughput by $14M in annualized revenue" beats "improved metrics through testing." If you can name the lift (11% incremental conversion) and the cost saving (18% less discount spend) in the same line, do it.

Avoid vague lines like "Worked on ML models" or "Used data to drive decisions." Data teams expect specifics: the method, the metric, and the magnitude.

Projects

Open-source and personal projects carry real weight in data science, especially for early-career candidates or anyone showing skills beyond their day job. Each entry here links to source and describes the work with concrete numbers. A reusable A/B test toolkit signals experimentation depth far more convincingly than a line on a skills list. If you're deciding how to present your repos, our note on how to put GitHub on a resume covers what to link and what to leave out.

Skills

Grouped by category: Languages, Machine Learning, Experimentation, Data & Infrastructure, Visualization. This is clearer than a flat keyword dump, and it mirrors how data science job descriptions split their requirements. Note the dedicated Experimentation group. For many roles (especially at product companies), A/B testing and causal inference are the differentiator, so they get their own line instead of being buried under "machine learning."

Education

For data science, education stays relevant a bit longer than in software, since many roles expect a quantitative graduate degree. Listing relevant coursework (Experimental Design, Bayesian Inference) is worth keeping for the first few years. Senior candidates can trim each entry to institution, degree, and dates.

What recruiters actually look for

Three things, in order:

Impact framed in business terms. Revenue, retention, latency, error reduction. The model is a means; the outcome is what gets you the interview.
Evidence of rigor. A/B testing, calibration, power analysis, variance reduction. These signal you know that a model in a notebook is not a model in production.
A stack that matches the job. Python and SQL are table stakes. The specific libraries and infrastructure should line up with the posting, which is exactly what the categorized skills section and bold_keywords make easy to adjust.

Picking a theme

The example uses engineeringresumes, a dense single-column layout based on the format recommended by r/EngineeringResumes. It works well for technical resumes because it maximizes content per page and parses cleanly.

If you want a bit more whitespace, engineeringclassic trades density for readability, which suits senior roles. Classic, sb2nov, and moderncv are solid general-purpose and academic options, useful if you're applying to research-oriented positions. Switching themes means changing one line of YAML; the content never moves.

This is the same approach as our software engineer resume example: one structured file, multiple layouts, full version control. Treating your resume as code means you can diff two variants, branch for a research role versus a product role, and reproduce the exact PDF months later.

Your resume deserves the same rigor you apply to a production model: validated input, reproducible output, and no black boxes between you and the result.

Build your data scientist resume at rendercv.com

Data Scientist Resume Example