Applied Scientist

An applied scientist builds and tests scientific methods — usually machine learning, statistics or optimisation — so they can be used in a real product or business decision. The job sits between research and engineering: you read and adapt published methods, prototype them on the employer's own data, measure whether they actually work, and then help turn the successful ones into something that runs reliably in production. Most applied scientists work in tech companies, e-commerce, finance, pharma or research-heavy consultancies, serving product teams and internal decision-makers rather than publishing for its own sake.

Starting salary: £45k–£70k
Live graduate jobs: 13 open now
Main way in: Direct application to advertised graduate or junior applied…
They usually ask for: A quantitative degree

Approximate graduate salary

Entry-level applied scientist roles in the UK typically start somewhere around £45,000–70,000, though this is a broad approximation and varies a great deal. Large technology and quantitative finance employers in London sit at or above the top of that range and often add equity or bonuses; smaller firms, startups, research institutes and roles outside London can start considerably lower, closer to general data science pay. PhD-entry roles generally start higher than MSc-entry ones.

What you'd actually do

Writing code — overwhelmingly Python, often with libraries such as PyTorch, scikit-learn or pandas — to prototype a model or method and see whether it beats whatever the business is currently doing
Cleaning and exploring messy internal data: working out what a column actually means, why half of it is missing, and whether the dataset can support the question being asked at all
Designing and running experiments, including offline evaluation (testing on historical data) and A/B tests (showing two versions to real users and comparing results)
Reading recent papers or technical blogs on a problem, and deciding which approach is worth the effort of adapting to the employer's data and constraints

Show 3 more

Writing up findings in a document with charts and clear recommendations, then defending the conclusions to engineers, product managers or non-technical stakeholders who want to know what to build
Working with software or ML engineers to get a prototype into production — handling latency limits, retraining schedules, monitoring for model drift
Code review, sprint planning and other normal software-team rituals; in many teams applied scientists are embedded in engineering squads

How graduates get in

Direct application to advertised graduate or junior applied scientist posts in large tech, e-commerce and financial firms — these are usually open to strong Master's and PhD candidates and are genuinely competitive
PhD route: in a lot of the market, 'Applied Scientist' is effectively a PhD-level title, particularly in machine learning, statistics, physics, computational biology or operations research. Many teams hire almost exclusively from PhDs, so a graduate with a bachelor's may find fewer doors open under this exact title
Master's conversion or specialist MSc (machine learning, data science, computational statistics, AI, bioinformatics) — a very common route for graduates whose first degree was in a related but not identical subject
Internships and summer research placements during a PhD or MSc — these are one of the most reliable pipelines, and converting an internship into an offer is common

Show 2 more

Sideways move after a year or two as a data scientist, quantitative analyst, research assistant or ML engineer — unusual to be the first job out of a bachelor's degree, very normal as a second job
Industrial PhD or CDT (Centre for Doctoral Training) programmes with an industry partner, and Innovate UK / KTP-style knowledge transfer placements, which embed researchers in a company for a fixed project

What employers ask for

A quantitative degree: computer science, maths, statistics, physics, engineering, economics, computational biology or similar. The subject genuinely matters here — the maths content is doing real work in the job, unlike many graduate roles where any degree is accepted
Postgraduate study is the norm rather than the exception. Many adverts ask for a PhD, or an MSc plus demonstrable research or industry experience. Some teams will take a very strong bachelor's graduate, but it is not the typical entry point
Strong coding ability in Python, usually tested in interview. Some employers also expect SQL, and some want C++ or Scala depending on the production stack
Solid grounding in probability, linear algebra and statistical inference — expect to be asked to derive or reason about a method, not just call a library function

Show 2 more

Evidence you can do original work: publications, a dissertation, a well-documented personal project, Kaggle-style competition results, or an internship where you shipped something. Which of these counts varies a lot by employer
Grades: a 2:1 or first is the usual bar for the first degree, though strong postgraduate work and research output can offset a lower undergraduate classification

Skills that matter

Experimental design and causal reasoning

Most of the job is answering 'did this actually help?', which means knowing about control groups, confounders, sample size and when a promising offline result will not survive contact with real users.

Production-quality Python

A notebook that only runs on your laptop is of limited use; applied scientists are expected to write code that engineers can read, test and deploy.

Statistical and ML fundamentals

You will regularly have to choose between methods, diagnose why a model is failing, and explain why a tempting shortcut is statistically invalid.

Problem framing

Business stakeholders arrive with a vague goal like 'reduce cancellations', and much of the value you add is converting that into a well-posed, measurable modelling problem — or saying honestly that the data cannot answer it.

Technical writing

Findings travel through written documents and short decks; work that cannot be explained to a product manager in a page tends not to get adopted.

Knowing when to stop

Research instincts push towards the elegant solution, but in industry a simple baseline shipped this quarter usually beats a sophisticated model that never launches.

Putting these on a CV? We counted which skills 12,914 live graduate adverts ask for, and what to do with the answer — part of our application advice.

Where it leads

Entry level (applied scientist I / junior / associate): you work on a scoped problem handed to you, with a senior scientist reviewing your approach and helping you navigate the data.
Mid-level: you own a problem area end to end, choose the method yourself, and are trusted to tell stakeholders when an idea is not worth pursuing. Reaching this typically takes a couple of years, but timelines vary widely by employer and how much the team ships.
Senior applied scientist: you set the technical direction for a workstream, mentor juniors, and are involved in deciding which problems the team takes on at all.
The path then usually forks. One branch is the individual-contributor track — principal or staff scientist, a deep technical specialist influencing work across several teams. The other is management — leading a science team, then a department. Neither is universally better paid or more senior; it depends on the employer.
Common exits: into data science or ML engineering leadership, into quantitative research in finance, into a startup as a founding technical hire, or back into academia or a research institute, though the return to academia gets harder the longer you stay in industry.

What people get wrong

“It is the same as a research scientist, just with a different label.”

At most employers the research scientist role is judged on novel contributions and publications, while the applied scientist is judged on whether something shipped and moved a metric. There is overlap, and some companies use the titles loosely, but the applied role is usually closer to the product and closer to engineering.

“You spend your days inventing new algorithms.”

A large share of the week is data cleaning, evaluation infrastructure, meetings about requirements, and getting an existing well-understood method to work on awkward real data. Genuinely novel method development happens, but it is a minority of the time in most teams.

“If your model is good enough, it gets used.”

Plenty of technically excellent work never launches because of latency limits, data privacy constraints, cost, or simply because nobody was persuaded. Influence and clear written communication determine adoption as much as model accuracy does.

“You need a PhD, full stop.”

It is the dominant route and many adverts ask for it, but not all. Some employers hire MSc graduates or promote strong data scientists and engineers into the title. The requirement is much firmer in big-tech research-adjacent teams than in smaller companies, where 'applied scientist' and 'data scientist' can describe near-identical work.

Where this varies

The title means noticeably different things in different places. In large technology and e-commerce companies it is a well-defined, PhD-heavy career ladder parallel to software engineering, focused on machine learning for a specific product surface such as search, recommendations, pricing or forecasting. In quantitative finance and insurance it shades into quantitative research, with more time on time-series modelling and risk. In pharma, biotech and some manufacturing or energy firms it can mean computational or laboratory-adjacent science — simulation, computational chemistry, bioinformatics, materials modelling — where the relevant background is a domain PhD rather than machine learning. In smaller companies and consultancies the title is often used interchangeably with data scientist and the work is broader and less specialised. Read the advert's required skills rather than the title, and check whether the role is embedded in an engineering team or sits in a separate research group, because that strongly shapes the daily work. Geographically, the highest concentration of these roles is in London, Cambridge and Edinburgh, with pharma-flavoured variants around Cambridge and the Oxford–Stevenage corridor.

See all 13 live Applied Scientist jobsFrom employers' own careers sites, updated daily Tailor your CV for this roleFree with an account, 3 a month

General guidance about the role across the UK market, not about any specific employer. Entry routes and requirements vary — always check the individual job advert.