Principal Engineer, Data & Agentic AI Engineering

GSK R&D and Digital & Tech's collective goal is to deliver GSK impact, including the acceleration of the discovery and development of medicines and vaccines to patients.

The Labs and CMC Tech function is central to this mission, driving the reimagining and business transformation of our global laboratory and CMC processes.

The Labs CMC Data, Agentic and Knowledge Engineering (DAKE) group architects the data, agentic and knowledge foundations upon which critical R&D processes and scientific workflows depend.

Reporting into the Senior Data and Agentic Engineer, the Principal Engineer, Data & Agentic AI Engineering, independently owns a specific Labs & CMC data domain.

The role designs pipelines for complex scientific use cases, owns the quality of what digital workers deliver in that domain, and keeps its context accurate and audit-ready.

Key Responsibilities

Domain Level Data and Agentic Engineering

• Design pipelines for complex Labs & CMC use cases and direct digital workers to build and operate them, with clear, defensible architectural trade-offs.
• Own the quality of agent-generated work in your scientific domain: review, test and approve it within the remit our operating model sets, and escalate what falls outside it to senior engineering leadership.
• Improve how digital workers operate in your domain by refining the frameworks, tools and evaluations they use, iterating on reliability signals so manual intervention falls measurably over time.

Context & Semantics

• Own the context for your Labs & CMC domain in the governed context layer: definitions, contracts, lineage and governance documentation that humans and digital workers rely on.
• Contribute reusable components to the broader Labs and CMC context layer, such as metadata enrichment, classification or lineage capture, that other scientific domains can adopt.

Data Quality & Governance

• Build an end-to-end data quality framework for your domain, encoding standards directly into the context layer with lineage tracking and governance documentation that passes audit.
• Apply privacy, IP, and consent controls in AI data pipelines, and manage sensitive R&D data appropriately across the data lifecycle.

AI Data Infrastructure & Lifecycle

• Design and build feature stores, vector databases or retrieval infrastructure for production AI systems within Labs & CMC, with data contracts that give each AI consumer consistent, quality-assured inputs.
• Manage training datasets, evaluation sets and data versioning for AI projects in production and contribute to evaluation dataset design with a clear view of how data decisions affect scientific model quality.

Platform & Infrastructure

• Build cloud data infrastructure independently using infrastructure as code and contribute reusable components to the team's shared data platform to reduce setup time for new Labs and CMC projects.
• Design observability into pipelines from the start, and resolve infrastructure issues with minimal disruption to scientific consumers. Own incident response for your domain, with clear post-incident analysis and follow-through.
• Raise team technical quality through code review and documentation, working closely with senior engineers and domain scientists to translate ambiguous requirements into well-scoped designs.

Basic Qualifications

• 6+ years of experience building and operating production data pipelines and data models.
• Deep SQL and Python, with strong hands-on expertise in modern warehouses or lakehouses and in orchestration and transformation frameworks, such as Databricks, BigQuery, Airflow, dbt or Spark.
• Experience designing and operating at least one AI agent or LLM-powered tool inside a data workflow, and fluent, daily use of AI coding assistants and agents with sound judgement on what to accept, change or reject.
• Experience designing and building feature stores, vector databases or retrieval infrastructure for production AI systems, including data contracts for at least one AI consumer.
• Experience managing training datasets, evaluation sets and data versioning for at least one AI project in production.
• A track record of building data quality frameworks and lineage for a domain, and of applying privacy, consent and compliance controls in AI data pipelines.
• Experience building cloud data infrastructure independently with infrastructure as code, with observability designed in from the start.

Preferred Experience

• Led the design and delivery of a component of a larger data platform.
• Experience with agentic systems in production: multi-step tool use, evaluations, guardrails and observability applied to operational data problems.
• Experience with quality frameworks and with catalogue, metadata or lineage tooling.
• Experience in regulated or high-stakes data environments, such as life sciences, labs, CMC, or healthcare, where strict data integrity is a mandate