Overview
We are seeking an experienced Data Scientist to join a multi-disciplinary team working on a UK public sector project focused on developing and deploying modelling and forecasting capabilities. The successful contractor will translate business requirements into rigorous modelling problems, build and evaluate time series and machine learning solutions, and assess the accuracy of large language model outputs. Collaboration with both technical and non-technical stakeholders is essential, along with effective communication of modelling approaches and results throughout the project lifecycle.
Responsibilities
- Translate requirements from analysts and business stakeholders into statistically or machine learning addressable problems.
- Design, build, and validate time series forecasting models using statistical methods and machine learning approaches.
- Apply supervised learning techniques, incorporating active learning strategies for model improvement.
- Conduct feature engineering to enhance the performance of forecasting and supervised learning models.
- Evaluate outputs of large language models using prompt engineering and retrieval-augmented generation techniques.
- Communicate modelling approaches to stakeholders and present model results iteratively throughout development.
- Maintain Python and SQL code under version control, adhering to DevOps practices for collaboration and deployment.
Requirements
- Proven experience in time series analysis, including both statistical and machine learning forecasting methods.
- Demonstrable expertise in supervised learning approaches, with active learning experience being an advantage.
- Ability to articulate business requirements into solvable modelling problems and communicate modelling outcomes effectively.
- Experience in evaluating large language models using techniques like prompt engineering and metadata support.
- Strong proficiency in Python and SQL programming.
- Experience with version control and DevOps methodologies for development and deployment.
- Capability to handle both structured and unstructured data sources.