Data Scientist III: NLP/LLM for IP Insights (Flexible Hours)

LexisNexis IP; United Kingdom; Hybrid; GBP 75,000 - 110,000; Full time

Benefits offered by this job

Health insurance

Pension scheme

Learning & development resources

Flexible working hours

Job summary

LexisNexis Intellectual Property (LNIP) is seeking a Data Scientist III to shape NLP/LLM-driven features for Protégé in PatentSight. You will design agentic workflows for multi-step IP tasks, build retrieval strategies, and evaluate model performance on large patent datasets.

Collaboration with engineers and IP experts will translate research into customer-ready solutions. The role emphasizes experimentation, rigorous validation, and delivering measurable business impact within a flexible,

Qualifications

• Degree in a quantitative or technical field (Statistics, CS, Mathematics, Data Science).
• Strong ML/NLP experience with LLM-based modeling and experimentation.

Responsibilities

• Develop NLP/LLM approaches and agentic workflows for IP tasks.
• Design retrieval strategies using hybrid search and evaluate relevance.
• Analyze large IP datasets to improve model performance and benchmarks.
• Translate results into product recommendations and business impact.
• Collaborate with product, IP experts, and engineers to align solutions.

Skills

NLP

LLMs

Python

Prompt engineering

RAG

Embeddings

Vector databases

Experimental design

Model evaluation

Education

Degree in Statistics/Computer Science/Data Science

Tools

LangChain

LangGraph

AutoGen

Job description

LexisNexis Intellectual Property (LNIP) is seeking a Data Scientist III to shape NLP/LLM-driven features for Protégé in PatentSight. You will design agentic workflows for multi-step IP tasks, build retrieval strategies, and evaluate model performance on large patent datasets.

Collaboration with engineers and IP experts will translate research into customer-ready solutions. The role emphasizes experimentation, rigorous validation, and delivering measurable business impact within a flexible,