Senior Analyst / Data Scientist

*Please be aware although we are a remote organisation we do require candidates to reside in the UK.*

Salary: £70,000 - £90,000

Workpattern: Full time 

Reporting to: Head of Analytics

#LI-Remote

About us

ProblemShared is a practitioner-led, CQC-regulated digital mind-health provider. We assess and treat adults and young people with ADHD, autism, and specific learning differences, and we partner with NHS trusts, integrated care boards, universities, insurers, and private healthcare providers. The company was founded by clinicians who watched the NHS waitlist problem from inside an A&E and decided to do something about it.

Our data mission: We believe “data for good” — that good data work transforms how neurodivergent people understand themselves, how practitioners sharpen their craft, and how teams design better support. The stack we work in: Fivetran → Databricks (Unity Catalog) → dbt → semantic layer → Dashboards and R/Quarto/jupyter for partner-grade reporting. Mature in some places, still being built in others. You’d help us close the gaps.

What makes us different

• Tool-agnostic, outcome-driven. We care more about a candidate’s mental agility and generalisable knowledge than tool-specific experience.

• Neurodiversity-welcoming. Potential over conformity. A meaningful share of our colleagues are neurodivergent. Adjustments to the process are normal, you don’t have to explain why.

• Fully remote, flexibly. Work where you do your best thinking. If you’re in London some of us meet up for coworking once a week.

Why this role matters
Our older reporting mechanisms work, but frequently through the manual toil of our office staff and some hard squinting over the results by our senior leaders. We have built a data platform to replace all of this. We are ingesting and synthesizing all of our organisations data, and we have begun rolling out automated dashboards and reports to close the last mile between the data and the decision. This role sits at that pointy end, figuring out what the clinical and business stakeholders need to see so they can act, and then making sure they can see it.

What you will do

• Engage internal partners (clinical, ops, commercial, marketing, finance, HR) to surface the decisions behind the request, then build the visual data communications that support them. You can expect 2-3 new requests each week.

• In the first instance you will answer these requests directly. Visualising and responding to the stakeholder through whatever tools get the insights to them fastest.

• Where the decisions are recurring, you will work with our team to ship reliable and well governed dashboards or reports, democratising access to the insights so you don’t have to answer the question again every month. These are built in code, and peer reviewed.

• You’ll push back when the question is the wrong question. Forecasting before the data supports it is hubris; small-sample inference is the bar. Tell us when not to build a dashboard.

• You’ll work with our data engineers and software engineers to ensure that data quality issues or ommissions get resolved upstream, not in your dashboard. We’re building reproducible flows, not a house of cards.

What good looks like at 90 days

• Two recurring partner-grade reports run from the platform with no manual stitching, reviewed and signed off.

• One of your peers has shipped visuals that you reviewed, and your expertise helped them improve it.

• You’ve pushed back on at least one stakeholder request that was asking the wrong question.

• Data quality issues you’re discovering are highlighted in your visuals (not concealed) and are flowing to our engineers via the data governance register.

What you bring – Must have

• Fluent in SQL on a modern cloud warehouse. Strong on grain, joins, dedup, and small-sample reasoning.

• Statistical literacy you can articulate — signal vs noise, significance, when a sample is too small to forecast from. Informed opinions on which models you would use and why.

• A track record of visual data communications that were understood and acted upon without you in the room. Evidence informed decisions is a legacy you leave in your wake, not advice you have to re-explain.

• Experience shipping EDA or reports or dashboards with at least two different visualisation tools, with a view on what work each is optimised for.

• Comfortable working with stakeholders who don’t know what they need. The requirements emerge from discussion and feedback on your prototypes, not a static brief.

We know nobody ticks every box. If you have most of these, please apply.

What you bring – Nice to have

• Python or R at comfortable-reading level or better.

• dbt — consuming the semantic layer is enough; authoring is a plus.

• Knowledge of reproduceable research principles, and an appropriate tool for doing it ( Quarto / Rmarkdown / Jupyter )

• Experience in a highly regulated industry working with sensitive data.

• Public work we can look at — GitHub, a blog post, a talk, chart samples. Optional, not expected.

• Any knowledge or interest in psychology, psychometrics or neurodiversity.

Our 2026 priority is information legibility and decision support; the heavy statistics and ML problems on the horizon can wait their turn. You’ll collaborate with our other analysts and engineers, and mentor them where your skills allow. They will mentor you in return where their skills allow.

How we work
You’ll work with pseudonymised clinical and operational data. Information governance is part of how you design analyses, not a step bolted on at the end. Analyses live in version control. Dashboards are built in code. Nothing ships from a notebook straight to a stakeholder without peer review.

We use AI coding tools only when they’re additive and always with human oversight. As part of the central data team, you’ll help set the standards the rest of the company follows — so we expect you to know when to lean into LLMs and when to distrust them.

We hold a particular stance on how we represent the people in our data. Our patients are neurodivergent, and our framing of them is fit, not fault — dignity precedes identity, recognition over categorisation, the person is the primary recipient of anything we say about them. If those phrases land for you, we’ll probably get on.

Async is the default. Deep-work blocks are real and protected. Slack response cadence is hours, not minutes. Meeting load averages around two hours a day. Stakeholder feedback is an important part of your design process, but your deep work is what gets the next prototype shipped.

What we offer you

• Excellent salary

• Annual Performance related bonus (Discretionary)

• Company Pension Scheme

• 30 days annual leave + public holidays + the option to buy and sell additional leave, & extended leave options such as sabbatical leave

• Private health insurance

• Blue Light card / discounts

• Enhanced family friendly policies

• Flexible working

• All company events and in-person team meet ups

• Access to a range of wellbeing activities

• Access to development / training opportunities to support your career ambition

• One volunteering day per year

Our Recruitment Process and Next Steps

1. Short call (30 min) Two-way interview; we don’t want to waste your time. We’ll describe the role so you can determine your own interest. There will be some statistical reasoning and data visualisation questions.

2. Online Programming Test (15-60 minutes). Just SQL at the querying layer. No hard algorithms – we just need to know you find easy programming questions easy (without an AI).

3. 2 Part Final Interview (60min/30min) – Technical conversation with the team, followed by a values conversation with the Chief Data Officer. Your choice whether these are back to back or on separate days – whatever is best for you.

We reply within seven days, yes or no.

If you’d like adjustments at any point in the process — written instead of live, extra time, breaks, captions, a quiet room, the questions in advance, anything else — tell us. Many of our team are neurodivergent; we expect to adapt the process to the candidate, not the other way around. No diagnosis or explanation needed.

All applicants welcome

No matter who you are, where you’re from, who you love, follow in faith, disability status, ethnicity or the gender you identify with, you’re welcome at ProblemShared.

We particularly welcome applications from autistic, ADHD, and other neurodivergent candidates — many of our own team are neurodivergent. You wouldn’t have to mask to do meaningful work here.