Hire data scientists who forecast what comes next and prove what worked

Hire data scientists tested on a modeling task before they reach your shortlist. We present matched candidates within 7 days of your request.

Abstract 3D illustration of small white spheres scattered along a smooth curved ribbon, one sphere in orangeAbstract 3D illustration of small white spheres scattered along a smooth curved ribbon, one sphere in orange

WHICH ROLE

Do you need a data scientist, an analyst or an engineer?

The roles that sit close to your data

Bring in a data scientist when your question is about the future or about cause: which customers will leave, what demand looks like next quarter, whether the new pricing really moved revenue. You interview the candidates we present, and the one you choose joins your team and works in your data.

Need a finished model more than a new team member? Then you are looking for machine learning as a project, which starts from an agreed scope.

Reports on the past call for an analyst

If the question is what happened last quarter, or why a dashboard dipped, you want a data analyst. Paying a data scientist to build reports is paying for skills you won't use.

Messy data needs an engineer first

A model can't be better than the tables under it. When your data sits in several systems and nobody trusts last week's numbers, start with data engineers who prepare the data, then bring in a data scientist to model it.

LLM features belong to AI engineers

Chat assistants, search over your documents and agents inside a product are mostly engineering on top of existing models, which makes them a job for an AI engineer. A data scientist is the hire when the model has to learn from your own numbers.

A model in production needs code around it

Even a good model needs an API, a scheduled job or a production pipeline before anyone uses it. Add a Python developer for that part, and the data scientist stays on the modeling.

Abstract 3D illustration of white block stacks side by side, one capped in orange, the roles comparedAbstract 3D illustration of white block stacks side by side, one capped in orange, the roles compared

THE WORK

What your data scientist will work on

The work starts with a question a report can't answer: who is likely to churn, how many orders to plan for, whether a new onboarding flow moved retention or the season did. A data scientist turns that into a machine learning model or a clean experiment, and tells you how far to trust it.

Most of it happens in Python, with R where your team's statistics already live, and SQL against your warehouse. Notebooks suit a question that is still open. Once a model has to run every day, it belongs in tested code in your repository.

Predictive modeling and ML

Forecasting, scoring and classification for churn, demand, credit risk or fraud, with scikit-learn and XGBoost on tabular data, and PyTorch or TensorFlow when deep learning earns its cost.

On tabular business data, gradient boosting often beats a neural network and is easier to explain to the people who act on its scores.

Experiments and statistics

Sample size set from the smallest lift worth detecting before the test starts, and no peeking at week one to call a winner.

When a clean test isn't possible, regression in statsmodels or a difference-in-differences comparison gets as close to cause as the data allows.

Feature engineering in Python

This is where most of the accuracy is won. Data gets joined and cleaned in SQL and pandas, then shaped into the signals a model learns from.

When the data outgrows one machine, PySpark on Databricks runs the feature work at cluster scale.

Validation and monitoring

A forecast gets tested on months it never saw, inputs get checked for leakage, and the model has to beat a simple baseline before anyone relies on it.

SHAP values show which inputs drive each prediction. After launch, the live model is scored against real outcomes as they arrive and checked for drift.

Bring us the question and the data you have

Tell us what you want to predict or test, the seniority you need and where your data lives. We'll come back with matched candidates.

HOW WE ASSESS

How do we assess a data scientist?

Three steps before your shortlist, and a modeling task you can ask about

Every candidate on your shortlist has passed three steps before you see a name: a technical skills assessment, a soft-skills evaluation and an analysis of their job history. The role decides what the first step asks.

The modeling task

For a data scientist, the technical assessment covers statistics and experiment design, A/B tests included, and SQL and Python on real data. Then comes a modeling task. A CV lists the libraries someone has used. The task shows whether they can tell a model that works from one that only looks good. We look for what a polished accuracy score can hide: data leakage, a validation split that fits the data (time-based for a forecast) and a comparison with a simple baseline. The candidate also has to explain the result, because a model nobody on your side understands won't get used.

Live coding and background checks

For some roles we add a live coding interview, where the candidate writes code and talks through each choice. A background check that confirms the CV is genuine runs when a client asks for one.

What you can check yourself

When you are bringing data scientists into your team, you can ask about the modeling task before you interview the candidate, and we share the task output on request. You still interview each person and decide who joins. Our technical interview, and live coding where the role has it, can be part of your own selection, with full visibility into both.

Abstract 3D illustration of white blocks weighed on a balance scale, with one block in orangeAbstract 3D illustration of white blocks weighed on a balance scale, with one block in orange

TECHNOLOGIES

Tools our data scientists work with

Python

R

SQL

pandas

scikit-learn

XGBoost

statsmodels

PyTorch

TensorFlow

MLflow

PySpark

Databricks

Snowflake

BigQuery

Redshift

AWS

BY THE NUMBERS

7 years

Average experience, company-wide

80%+

Senior-level or above, company-wide

50+

Projects delivered by our team

PROCESS

How staff augmentation runs, step by step

The durations below are usual times. What we commit to is presenting candidates within 7 days of your request, and nobody joins until you have interviewed them. Hiring other roles too? See the complete roster of engineers.

  1. STEP 1

    Agree the role and the question

    We agree the role, seniority and stack with you, plus the question the data scientist is there to answer and the data sources they will use.

    1 day

  2. STEP 2

    Shortlist assessed candidates

    We shortlist people who have passed our technical assessment, soft-skills evaluation and job-history analysis, modeling task included.

    1–2 days

  3. STEP 3

    Interview and choose

    You interview the people you pick, turn down anyone who doesn't fit, and decide who joins.

    1 day

  4. STEP 4

    Sign, set up access and start

    We handle contracts, the NDA and security setup, and your data scientist joins your team through the accounts you issue.

    1–2 days

HOW WE WORK

Engagement models and terms

Staff augmentation

Our engineers join your team and work in your process and tools. You manage them; we source, vet and contract them.

Hourly rate per engineer by role and seniority, invoiced monthly.

  • If an engineer isn't the right fit, we replace them within 14 days at no extra cost.
  • Knowledge transfer to the replacement engineer isn't billed.
  • There is no free trial period.
  • 30 days' notice to take an engineer off or end the engagement, with no minimum term beyond it.
  • Out-of-hours on-call only if agreed with you; we set its hours and billing together.
  • After two years, moving an engineer onto your own staff is something we can discuss.

Managed team

An agreed team capacity that we manage. We coordinate delivery and handle the team's make-up. You set the priorities and own the backlog.

Per-engineer rates, or a fixed monthly fee for the agreed capacity.

  • We handle replacements ourselves, and the team keeps its agreed capacity and pace.
  • Engineers work full-time for your team only.

Project delivery

We deliver an agreed scope end to end: plan, team, milestones, QA and release. You make the product decisions and accept each milestone.

Fixed price per milestone after discovery, or time and materials against an estimate.

  • Scope changes go through change requests.
  • On fixed-price projects, a warranty period agreed in the contract, usually 30–60 days.

Whichever model you choose

  • We reply to an inquiry from the contact form within 24 hours.
  • We sign an NDA before work starts, and IP passes to you on payment.

GROUND RULES

What to settle before the work starts

Abstract 3D illustration of white blocks behind a gate with an orange key in its lock, access settled before work startsAbstract 3D illustration of white blocks behind a gate with an orange key in its lock, access settled before work starts

Access through accounts you issue

A data scientist reaches your data, your warehouse and your models only through named accounts you issue, scoped to the role, and every one of them is removed at exit.

Hours that overlap with yours

CET/CEST business hours are the default, so EU and UK teams get the full working day. For a US team we shift the schedule for 4+ hours of daily overlap with the West Coast and more with the East.

Inside your tools and process

Under staff augmentation, a data scientist joins your stand-ups, takes tickets from your tracker and commits to your repositories. Their notebooks and code usually live there too.

Data handling agreed in writing

Where we handle client data, a data processing agreement is part of the engagement. Say which datasets are in scope and which fields, such as personal data, to mask before work starts.

Start with what you want to predict

"Ask any model what it had to beat. A good data scientist names the baseline before anyone mentions accuracy."

Dzmitry Kuchynski, Chief Operating Officer at Octalance

Dzmitry Kuchynski, Chief Operating Officer, Octalance

FAQ

Questions to settle before you sign

What should I ask a data scientist about their last model?

Ask how they kept future information out of the training data, and how they split the data when the model forecasts over time. Ask what simple baseline it had to beat, and by how much. For a rare outcome such as fraud, ask which metric they chose and why plain accuracy would have misled them. Then ask how they fixed the length of their last A/B test before it started. Someone who has shipped models answers these with examples; a vague answer is worth a second question.

How is hiring a data scientist priced?

The bill moves with seniority and with how many data scientists you take on. Under staff augmentation each one is billed at an hourly rate, so send us the role and the seniority and we will come back with a figure. The notice period and the price types of the other two models are under Engagement models and terms on this page.

We have no data science lead. Who directs the work?

Under staff augmentation, your own tech lead still sets the priorities. For work nobody on your side has done, ask for a lead-level data scientist to lead the technical side: framing the business question as something you can test, setting the bar a model must clear before it goes live, and saying plainly when your data can't answer the question. If you want the work run end to end, project delivery is the route, with a test set agreed with you before the build.

Who puts a model into production and watches it afterwards?

Watching a live model for drift and checking it after launch is work our data scientists take on. Deploying it is a separate job. If your team already ships Python services, they usually take the packaging; if not, plan an engineer beside the data scientist from the start. Leaving deployment to the end is a common reason a good model never gets used.

Who owns the work when the engagement ends?

IP passes to you on payment. The code usually sits in your own repositories, sometimes in ours, and transition-out at the end is routine. An NDA is signed before the start.

What data science work can you show us?

Two published cases. We rebuilt a logistics company's demand forecasting system on Databricks and Azure. The new forecast reads seasonal changes, economic indicators and competitor data alongside the shipment history, and our client reports its accuracy rising from 70% to 92%. For a travel company we built an ML model that predicts each user's revenue, applied inside a real-time discount pipeline. The discount pipeline ran beside the base one as an A/B test, and the comparison showed our client 32% more ticketing profit on that channel. Our other data-science work sits under client NDAs. Under staff augmentation, ask for a candidate's modeling-task output before you interview them.