AI/ML Research Engineer

We are seeking an AI/ML Research Engineer (NLP & Knowledge Graphs) to join our Applied AI Solutions team. This is a junior-to-mid level role: the ideal candidate has around 2–4 years of hands-on experience applying NLP and machine learning (or an MSc with strong research or internship experience), and wants to grow into a research-minded engineer who can take an idea from paper to working prototype.

You will research, prototype and evaluate ways of extracting structure and meaning from scholarly content – entities, relations, topics and emerging research areas – and of representing it in knowledge graphs and ontologies that can power better discovery, search and recommendation across Taylor & Francis. You will report to the Director of Applied AI Solutions, who in turn reports to the CTO.

The Applied AI Solutions team is a small, senior, research-led group within the CTO organisation. We are the part of Taylor & Francis that works on problems that are on nobody’s product roadmap yet – discovery, peer review, research integrity and making scholarly content machine-consumable – and we turn the most promising ideas into evidence, working prototypes and business cases that Engineering and the business can adopt.

What you’ll be doing:

• Work as part of a small, research-led team on a few strategic AI “bets” each year, each ending in a decision-grade artefact: a working prototype, the evidence behind it and a business case.
• Research, prototype and evaluate NLP approaches for extracting entities, relations and topics from scholarly articles, books and metadata.
• Help design, populate and evolve knowledge graphs, ontologies and taxonomies that represent Taylor & Francis content and the research landscape around it.
• Build and benchmark semantic and hybrid retrieval approaches, including embedding models and graph-enhanced retrieval-augmented generation (RAG).
• Contribute to the team’s technology radar by scouting, trialling and writing up new open-source libraries, models and tools for knowledge extraction and search.
• Build test sets, metrics and evaluation harnesses so that every claim we make about a model or tool is backed by evidence.
• Support the hand-over of graduating prototypes to Engineering teams, with clear documentation and a written definition of what “done” means.
• Present results clearly to technical and non-technical audiences, and contribute to papers, articles and conference submissions.
What we’re looking for:

Key Responsibilities:

• Develop NLP pipelines for tasks such as named-entity recognition, relation extraction, entity linking and disambiguation, topic modelling and classification over scholarly text, using transformer models and LLMs.
• Build, populate and query knowledge graphs in a graph database or triple store (e.g., Neo4j, RDF/SPARQL stores) and evaluate their quality and coverage.
• Support the design and expansion of ontologies, taxonomies and controlled vocabularies, working with Editorial and subject-matter experts.
• Detect and track emerging topics and research areas using clustering, topic modelling and embedding-based techniques.
• Evaluate embedding models, chunking strategies, rerankers and retrieval configurations for semantic search and RAG on scholarly content.
• Create gold-standard datasets and annotation guidelines, and define metrics (e.g., precision/recall/F1, nDCG) for reproducible evaluation.
• Design and run well-controlled experiments, tracking them with tools such as MLflow or Weights & Biases so results can be reproduced.
• Prototype rapidly using LLM APIs and open-source models for knowledge extraction and enrichment.
• Write clean, tested and documented Python code that other engineers can pick up and extend.
• Prepare hand-over documentation and support Engineering teams during the build phase of prototypes that graduate to production.
• Collaborate with Engineering, Product, Editorial and other business teams to turn open-ended problems into well-posed research questions.
• Assist with vendor, model and tool comparisons, including preparing benchmark results for internal and external audiences.
• Keep up with the research literature and relevant venues (e.g., ACL, EMNLP, SIGIR, ISWC/ESWC) and share what is relevant with the team.

Function: Information Technology
Experience Level: Mid-Senior Level