Data Scientist: NLP & AI for Consumer Insight

Reference
1113400

Branch
University of Nottingham

Start date
22/06/2026

Duration
3 Months

Hours and days
To be arranged by client - 37.5 hours/week for the first 12 weeks; in the final month, hours and days will be ad-hoc as needed and would not exceed 30 hours per week.

Dress code
Smart Casual

Rate of pay
-

Holiday rate
-

Role

We are recruiting a fixed-term Research Associate/Data Scientist for a 4-month feasibility project with Boots. The role will develop and pilot an AI workflow that turns qualitative consumer feedback (e.g. transcripts and open-text responses) into structured, decision-ready descriptors to support faster R&D screening decisions. You will work with an academic PI, a PhD researcher and Boots colleagues to build, test and document the workflow. The project will provide hands-on experience of applied NLP/AI in an industry-facing setting. 

Start Date: 22/06/26

End Date: 30/09/26 

Working Hours: To be arranged by client - 37.5 hours/week for the first 12 weeks; in the final month, hours and days will be ad-hoc as needed and would not exceed 30 hours per week.

Pay Rate: £20.00 per hour 

Holiday Pay Rate: £2.41 per hour 

Location: Online but can be on campus (Sutton Bonington) if preferred 

Dress Code: Smart casual  

Duties and responsibilities

• Build end-to-end text-to-descriptor pipeline that maps qualitative consumer language to latent affective/semantic space and aligns outputs to a domain lexicon (embedding-based and semantic similarity methods). 
• Implement and evaluate an affective mapping component (e.g. VAD-style continuous representation)and retrieve interpretable descriptors via nearest-neighbour search in a domain-specific lexicon. 
• Create a small annotated subset and benchmark semantic mapping quality and document error modes. 
• Design structuring approaches (including LLM-based methods where appropriate) to support embeddings and mapping pipelines 
• Run robustness testing of the pipeline (e.g. sensitivity to input phrasing, stability of embeddings, and variation across models/configurations). 
• Deliver a Boots pilot on real qualitative data and quantify time save dvs manual coding and assess quality trade-offs (e.g. interpretability, scalability) and write a concise pilot report.

Skills and experience

• Strong Python skills for NLP/data science (e.g., pandas, scikit-learn, PyTorch, text processing); reproducible workflow habits (Git). 
• Experience using transformer models (e.g. BERT, RoBERTa) for embeddings extraction and semantic similarity tasks. 
• Experience working with qualitative data or unstructured text, including building embedding-based NLP pipelines, including text mapping, neural regression, and cosine similarity / nearest neighbour retrieval. 
• Experience with statistical and clustering methodologies(e.g. clustering, similarity metrics, evaluation metrics). 
• Familiarity with LLM prompt engineering and structured outputs (as a supporting component) 
• Ability to design pragmatic validation/benchmarking approaches (gold standard subset, interannotator agreement concepts, error analysis, and evaluation of sematic mapping quality and overfitting/generalisation)
• Experience in GDPR, handling sensitive text data responsibly (data governance, anonymisation awareness; secure storage/processing). 
• Strong communication skills: ability to explain technical work to mixed academic/industry audiences and write clear documentation.

Location
Online but can be on campus (Sutton Bonington) if preferred

How to apply
Please apply here!