Principal D

About the Role

Lead end-to-end domain-specific content modeling and production NLP work for large-scale scientific text, owning model development, evaluation, and deployment. Build and operate high-volume enrichment pipelines that combine classical NLP, embeddings, and LLM-based approaches to deliver reliable research intelligence to downstream applications.

Works directly with LLMs and agentic AI, comparing prompt-based and fine-tuned approaches and operating high-volume LLM workloads in production.

Location: United Kingdom

Salary: GBP 59,100 - 84,633.33/year

Type: Full Time

Posted: Yesterday

Category: AI/ML & Data

Vibe Markers: 💻 Open Source, 🔥 Hot

Experience Level: Staff/Principal

Job Description

Role

Principal Data Scientist (NLP + Applied AI) responsible for owning domain-specific content modeling end-to-end for a large scientific corpus, including development, evaluation, and production deployment of NLP models and enrichment pipelines.

Key Responsibilities

• Design and build scalable NLP enrichment pipelines to extract entities, classifications, claims, and summaries from scientific full text.
• Evaluate and compare approaches (classical NLP, embedding-based retrieval, LLM prompting, fine-tuned smaller models) and select tools based on evaluation, cost, and operational tradeoffs.
• Own evaluation: create golden sets with SMEs/vendors, choose metrics, and balance speed, quality, and cost.
• Write production-quality Python, manage concurrency and cost for high-volume LLM workloads, and structure code for engineering and data science use.
• Collaborate with data engineers to orchestrate work using data pipeline tools (Airflow, Dagster) and design idempotent, retryable pipeline stages.
• Contribute to agentic AI application work where enriched corpus grounding and evaluation inform tool-using systems.
• Work directly with editors, product managers, and engineers to translate stakeholder needs into modeling priorities.

Requirements

• Deep, production-grade Python experience, including concurrency patterns and tradeoffs (asyncio vs threads vs queues).
• Strong NLP background across modern (LLMs, transformers, embeddings, retrieval) and classical (NER, sequence labeling, classification) approaches with hands-on evaluation experience.
• Habit of comparing approaches and defending choices with evaluations and cost estimates; ability to handle model performance drift.
• Track record of shipping systems (not just prototypes or papers) that deliver value to real users.

Nice to Have

• Experience with scientific or scholarly text.
• Familiarity with AWS (S3, Batch, Lambda, SageMaker) and Parquet or Iceberg data lake patterns.
• Experience running LLMs under realistic cost and latency budgets in production.
• Exposure to agentic AI applications: tool use, multi-step reasoning, guardrails, and trajectory evaluation.

Why this role

Work with one of the largest scientific corpora to apply modern AI and build systems that accelerate research, delivering measurable impact to researchers and products.

Salary

• Base salary range (as posted): 59,100.00 GBP to 84,633.33 GBP

Tech Stack

Python, Airflow, Dagster, AWS S3, AWS Batch, AWS Lambda, Amazon SageMaker, Parquet, Iceberg, LLMs, Transformers, Embeddings, Retrieval, NER, Sequence labeling, Classification

Skills

NLP, Model evaluation, Production ML, Production-quality Python, Concurrency and cost management, Data pipeline design, Pipeline reliability and idempotency, Experimentation and A/B evaluation, Stakeholder communication, Collaboration with data engineers, Agentic AI reasoning

Benefits

• Meeting-free Friday afternoons
• Continual learning and internal mobility
• Professional development
• Comprehensive benefits package
• Competitive compensation
• Reasonable accommodation
• Employee programming and community events