How is a data scientist CV different from a data analyst CV in the UK?

The line has moved in 2026. Data analysts answer business questions and rarely own production code; data scientists ship models that make automated decisions and stay on the hook for them after launch. On a CV, that means analysts lean on SQL, dashboards and stakeholder bullets; scientists lean on production models, MLOps, monitoring and model-risk work. If you are mid-career and you ship models, lead with the production work. If you sit between the two roles, position yourself as 'applied data scientist' or 'analytics engineer' and let the bullets show which way you lean.

Do I need a PhD to get hired as a data scientist in the UK?

No, and the requirement has softened markedly in the last few years. PhDs are still common in research-leaning roles (pricing, fraud research, deep learning) and at the largest banks, but most UK applied data science teams will hire on the strength of production work and commercial impact regardless of your degree. If you don't have a PhD, lean hard on shipped models, business outcomes and the breadth of your stakeholder work. If you do have one, push the thesis to a single line at the bottom unless it's directly relevant to the job.

Should I include Kaggle or open-source projects on my CV?

Only if they're substantial. A handful of finished competitions, a published package, or a notebook with thousands of stars can earn space on the CV. A long list of half-finished competitions hurts more than helps. The bar in 2026 is roughly: would a senior data scientist look at this and learn something? If yes, include it under a 'Selected open work' heading with a one-line outcome each. If you mostly do confidential commercial work, skip Kaggle entirely and let the production bullets carry the weight.

CV Example · Tech · UK 2026

Data Scientist CV Example UK

Data scientist CVs in the UK have gone through a useful correction. The 'Kaggle competitions and a Coursera certificate' template doesn't get interviews in 2026. After 12 years placing scientists into UK fintech, retail and consultancy teams, the strong CVs now look more like applied research portfolios: real production models, real business outcomes, and a clear stance on the trade-off between accuracy and shipping. Hiring managers want to know whether you can take a vague commercial question, frame it as a model, ship it into production, and own it after launch. Tooling lists are the easiest part of the CV to over-invest in and the easiest part to skim past.

By Alex · 12-year UK recruiter · Updated April 2026

Example header

Tom Bailey · Senior Data Scientist · 6 years · London / Hybrid

Personal statement / Professional summary

Applied data scientist with six years building production models for UK fintech and insurance. Specialise in fraud, credit risk and pricing, with hands-on experience taking models from notebook to live decisions on customer accounts. Best work last year: rebuilt the transaction-fraud model at a challenger bank, cutting false positives by 41% on £2.3bn monthly card volume while holding fraud capture flat. Comfortable on call for model performance and the awkward conversations when a model starts drifting.

Bullet point examples

Strong bullets follow the same shape: action verb, specific scope, quantified outcome. Use these as patterns, not as copy-paste templates — the numbers must be your own.

Lead scientist on transaction fraud, challenger bank

Rebuilt the production fraud model (gradient-boosted trees, real-time features) on £2.3bn monthly card volume, cutting false positives by 41% while holding fraud capture flat.
Designed the feature store and online inference path with platform engineering, getting end-to-end model latency under 80ms p99.
Set up daily drift and PSI monitoring with PagerDuty alerts; caught a feature pipeline regression in 4 hours that would historically have taken a week.

Credit risk and pricing

Built a probability-of-default model for unsecured lending, validated against 18 months of book performance, that improved Gini from 0.51 to 0.63 and unlocked an additional £18m of approved lending in the first year.
Co-led pricing analysis with commercial that informed a risk-based pricing rollout across three product lines, with documented sensitivity to regulator-relevant scenarios.

Production MLOps

Migrated 6 production models from ad-hoc Airflow jobs to a managed MLflow + SageMaker pipeline, cutting model release cycle from 6 weeks to 9 days.
Wrote the team's model risk documentation template, accepted by internal audit and now used for all new model deployments.

Stakeholder and commercial work

Translated regulator queries on the fraud model into plain-English memos for the head of compliance and the FCA-facing legal team across a 4-month review.
Ran a quarterly model performance review with commercial leads, including loss attribution and a written recommendation on retraining cadence.

Earlier role: scientist at retail analytics consultancy

Built a demand-forecasting model for a UK grocery client across 1,400 stores; reduced waste on fresh categories by an estimated £4.2m annualised.
Co-authored a customer segmentation framework adopted as the firm's standard for retail engagements.

Skills section — what to list

Mirror the skills exactly as they appear in target job ads. The ATS reads this section literally — synonyms hurt match scores.

Python (pandas, scikit-learn, PyTorch)SQL (advanced)Gradient-boosted trees (XGBoost, LightGBM)Time-series forecastingMLflowAWS SageMaker / GCP Vertex AIAirflow / DagsterFeature stores (Feast)Model monitoring and drift detectionBayesian and frequentist statisticsA/B and quasi-experimental designdbt and modern data stackGit and code reviewModel risk documentationStakeholder communication

Data Scientist-specific CV mistakes that get you binned

× Listing every algorithm you've read about. UK hiring managers in 2026 read 'XGBoost, Random Forests, SVM, KNN, Naive Bayes' as a list copied from a textbook.
× No production work. If every project is a notebook on Kaggle or a university dataset, you read as a graduate. Even one 'shipped to production' bullet changes the conversation.
× Skipping the business outcome. 'AUC of 0.87' is half a story. The other half is what changed for the business.
× Putting a long 'Education and certifications' block on page one. After your first job, push it to the bottom and protect the front for production work.
× Claiming you 'led' work that the engineering team actually shipped. Senior data science hiring managers will pull on that thread in 30 seconds at interview.

Common questions

How is a data scientist CV different from a data analyst CV in the UK?: The line has moved in 2026. Data analysts answer business questions and rarely own production code; data scientists ship models that make automated decisions and stay on the hook for them after launch. On a CV, that means analysts lean on SQL, dashboards and stakeholder bullets; scientists lean on production models, MLOps, monitoring and model-risk work. If you are mid-career and you ship models, lead with the production work. If you sit between the two roles, position yourself as 'applied data scientist' or 'analytics engineer' and let the bullets show which way you lean.
Do I need a PhD to get hired as a data scientist in the UK?: No, and the requirement has softened markedly in the last few years. PhDs are still common in research-leaning roles (pricing, fraud research, deep learning) and at the largest banks, but most UK applied data science teams will hire on the strength of production work and commercial impact regardless of your degree. If you don't have a PhD, lean hard on shipped models, business outcomes and the breadth of your stakeholder work. If you do have one, push the thesis to a single line at the bottom unless it's directly relevant to the job.
Should I include Kaggle or open-source projects on my CV?: Only if they're substantial. A handful of finished competitions, a published package, or a notebook with thousands of stars can earn space on the CV. A long list of half-finished competitions hurts more than helps. The bar in 2026 is roughly: would a senior data scientist look at this and learn something? If yes, include it under a 'Selected open work' heading with a one-line outcome each. If you mostly do confidential commercial work, skip Kaggle entirely and let the production bullets carry the weight.