Updated October 2026
A strong startup data scientist job description names which kind of data scientist you need, analytics, decision science or machine learning, and the decisions or product the hire will change in year one. It lists outcomes, keeps must-haves to six and is honest about how clean your data is.
Decide the flavour before you write a line. The title covers three different jobs, and a JD that asks for all three attracts a modeller who will be bored by the analysis you actually need.
Our running example is an illustrative Seed-stage language-learning app. The founder's draft said "build our recommendation engine", but the real questions were why users drop off in week two and whether the new paywall worked. That is analytics and decision science, not machine learning.
Copy-and-paste data scientist job description
About [Company]
[Company] builds [one sentence on the product] for [who the customer is]. We raised [round] from [investors] in [month, year] and have [number] people. Our data lives in [product analytics tool] and [warehouse or database].
The role
We're hiring our first data scientist, focused on [analytics / decision science / machine learning]. You'll report to [founder or Head of Product] and work with [product, growth or engineering lead] on [the decisions or product area you will change].
What you'll do in your first 6 months
- Answer [the questions founders ask most, e.g. why users drop off in week two] with analysis the team acts on.
- Design and read [number] experiments on [onboarding / pricing / paywall], with sample sizes agreed before launch.
- Run a weekly metrics review the founders use, with uncertainty shown rather than hidden.
- Fix the [number] data gaps that block analysis, working with [engineering lead].
- [ML flavour only] Ship a first [model] to production that beats a simple rule-based baseline, with monitoring.
What you'll bring
- You've done analysis or built models that changed a decision, and you can name it.
- Strong SQL and [Python / R].
- Solid statistics, especially [experiment design / causal inference / forecasting].
- You can explain uncertainty to a non-technical founder without jargon or false precision.
- You've worked with imperfect data and said clearly what it could support.
- [ML flavour only] You've put a model into production and kept it running.
Nice to have
- Experience in [industry or product type].
- Event tracking in [product analytics tool].
- Enough data engineering to build a simple pipeline.
- You've been the first data person in a team.
What we offer
- Salary: [£/$/€ X to Y], depending on experience.
- Equity: [X% to Y%] in [options / shares], [vesting schedule].
- [Remote / hybrid, X days a week in City].
- Stage: [Seed / Series A / Series B]. [Benefits].
For the language app, the founder deleted both ML lines and replaced "build our recommendation engine" with "explain week-two drop-off and test three fixes". The free job description builder handles edits like that from rough notes, and our data scientist interview questions follow the same flavours.
How a data scientist job description changes by stage
| Seed | Series A | Series B | |
|---|---|---|---|
| Flavour to hire first | Analytics, often a senior analyst | Analytics or decision science; ML only if the model is the product | Separate roles per flavour |
| Data they inherit | Events tracked, little else | A warehouse, ideally a data engineer | A data team with engineers |
| Who they work with | Founders directly | Product and growth leads | Their own manager, embedded in a team |
Benchmark salary and equity against current data for your market; the flavours price differently.
Scorecard: how you'll judge success
| Outcome | What good looks like at 12 months | How you'll test it in interviews |
|---|---|---|
| Decisions changed | Founders can name decisions made on their work | Ask which decision their work changed, and who made it |
| Experiment quality | Tests sized before launch, failures reported too | Give them a finished test with a borderline result |
| Communicating uncertainty | Founders understand the range and still act | A five-minute readout to the founder |
| Data foundations | Blocking gaps listed, the worst fixed | Ask what they would check first in our tracking |
| Models in production (ML only) | Live, monitored, beating baseline | Ask about a model that never shipped |
30-60-90 day plan
| Days | Focus | By the end they should have |
|---|---|---|
| 1 to 30 | Sit in on founder decisions, audit tracking | A ranked list of unanswered questions |
| 31 to 60 | Two analyses that change a decision; first experiment live | A pre-agreed readout date |
| 61 to 90 | Weekly metrics review; [ML: baseline model in shadow mode] | A review the founders rely on |
Mistakes founders make in data scientist job descriptions
Asking for all three flavours. Deep learning, A/B testing and Tableau in one must-have list describes three people.
Hiring a modeller before the data exists. If events are untracked and tables disagree, read our data engineer job description first.
Counting models built. Models that never reach production are the commonest waste in this role. Measure decisions changed and models live.
- If most of your six-month list is dashboards and ad hoc questions, retitle it product analyst or senior analyst.
- If nobody can tell you which events are tracked, hire or borrow a data engineer first.
- If an ML line survives, name the model and where it will run.
Where Funded.club fits
We've helped 500+ startups from Seed to Series D across North America, Europe and APAC. Our fee is fixed and agreed upfront, averaging 6 to 9% of salary. Illustratively, on a $105,000 hire a 20 to 25% contingency fee would be $21,000 to $26,250; ours is $7,500 (see pricing).
One dedicated recruiter runs the search, with first screened candidates within 7 days and 33 days on average from kick-off to hire. For the machine learning flavour, read our guide to hiring ML and AI engineers for a startup.
Frequently asked questions
What is the difference between a data analyst and a data scientist job description?
An analyst JD centres on reporting and business questions. A data scientist JD adds experiment design, statistical modelling or production models.
Should a startup's first data hire be a data scientist?
Usually not. A data engineer or a strong analyst comes first, because a data scientist needs reliable data. The exception is when a model is the product.
Should the job description name the flavour in the title?
Yes. Write "Data scientist, product analytics" or "Machine learning engineer" rather than the bare title, so the right specialists recognise themselves.
Worth a brief chat about your next hire? Book a free call.
Hiring after a funding round?
First candidates in 7 days. Fixed-fee, no commission.