Data Scientist

Data Scientist

💰 £400-495 per day pay Inside IR35 via umbrella. 3-6m month contract

📍 Fully remote (UK based)

** You must hold or be eligible for SC clearance.**

Company & role

This role sits with a leading Microsoft Solutions Partner delivering advanced data and AI projects into large public sector environments. They are building an AI powered forecasting assurance solution and need a Data Scientist to lead the design, development and evaluation of the models that sit at the heart of it.

The premise is genuinely interesting. You will build predictive models that generate independent challenge forecasts, compare them against existing financial forecasts, flag material variances, and produce explainable evidence that supports financial review and challenge. It is a blend of time series forecasting, statistical analysis and Generative AI, applied to a real problem where the output actually changes how decisions get made.

You will work alongside finance SMEs and data engineers, own the modelling end to end, and hand over to the engineering team for production build.

Why This Role Stands Out

You are leading the technical direction of the solution, not slotting into someone else's pipeline. The work spans classic forecasting and modern LLM driven extraction, so you get to combine both rather than pick a lane.

It is a well defined proof of concept through to MVP piece of work, which means clear scope and visible impact inside six months.

Fully remote, day rate contract with a strong problem to get your teeth into.

You are building explainability into the models from the start, linking forecast movements back to cost centres, activities and narrative evidence, which is where a lot of this work usually falls short.

Key Responsibilities

• Develop and evaluate time series forecasting models using historical spend data and current year actuals
• Build AI generated challenge forecasts to act as an independent comparator against existing finance forecasts
• Design variance detection and prioritisation algorithms to surface material forecast differences
• Apply LLMs to extract structured information from finance narratives, leadership reports and supporting documentation
• Engineer features from both structured financial data and unstructured text
• Build explainability approaches that connect forecast outputs to financial drivers and narrative evidence
• Evaluate model performance using forecasting metrics such as signed, absolute and weighted error
• Present findings and model insights to finance stakeholders and senior leadership
• Support knowledge transfer and handover to the engineering team for production implementation

Ideal Experience

• 5+ years in Data Science, Machine Learning or Predictive Analytics
• Strong time series forecasting background, with real experience evaluating forecast accuracy
• Solid statistical analysis, hypothesis testing and anomaly or variance analysis
• Hands on with LLMs (GPT, Azure OpenAI or equivalent), NLP, prompt engineering and information extraction
• Familiarity with explainable AI approaches
• Python essential. Pandas, NumPy and Scikit learn valued
• Experience with forecasting frameworks such as Prophet, ARIMA, Chronos or NeuralForecast is desirable
• Experience with Azure Machine Learning, Databricks or similar, plus SQL and Git
• Comfortable integrating structured and unstructured data, with Power BI or equivalent for visualisation
• Public sector, financial planning or forecasting and assurance experience is a strong advantage
• Confident engaging stakeholders and translating complex analytics into clear business insight