Research Engineer - Scientific Knowledge and Field Building (12 months FTC)

Based at our office in London, we're looking for a Research Engineer - Scientific Knowledge and Field Building. Help UK DRI understand what is known, where the evidence disagrees and what work could make the biggest difference next. You can learn more about our ambitions here.

You will develop the strategy and build software that makes these questions easier to investigate and act on. Starting with a focused Alzheimer’s disease research question, you will help scientists create and maintain an inspectable map of claims, evidence, disagreements, capabilities and possible routes towards a meaningful outcome. Over time, that work should connect published literature with selected, authorised unpublished research and the knowledge held across UK DRI laboratories.

We are looking for a technical leader who thinks big and cares about how science establishes knowledge and how institutions help people learn and work together. You might already be interested in discourse graphs, claims and evidence, outcome graphs, reproducibility, open science or new ways of organising research. You might have encountered the same problems while building search tools, scientific data systems or collaborative software. We value the ability to turn that curiosity and ability to think big into useful working systems.

You will be a founding member of the UK DRI AI initiative, working as a peer with an engineer building Lab Brains and scientific workflows. Together, you will connect what laboratories learn and influence what the institute can understand and do.

What you'll build

• A practical representation of scientific questions, scoped claims, supporting and opposing evidence, sources, methods and uncertainty. Keep exact source links, versions and the distinction between proposed and reviewed interpretations.

• Ingestion, extraction, reconciliation and update pipelines for an agreed scientific corpus, with interfaces that make it easy for researchers to inspect, qualify and correct the results.

• Views that reveal where findings agree, genuinely disagree or cannot yet be compared, without treating missing evidence as evidence against a claim.

• A first Alzheimer outcome map: work with scientists to describe what must become possible for a selected outcome, the alternative routes, unresolved assumptions, missing capabilities and next decisions.

• Connections between that map and reviewed institutional capabilities, datasets and developing research, making potential collaborations or shared bottlenecks specific enough for people to evaluate.

• Decision dossiers and feedback tools that record what was considered, what an accountable owner chose, what was tried and how subsequent evidence changed the picture.

Choose data models, search and graph technologies according to the task. Learn from communities such as Discourse Graphs and existing outcome-planning approaches, compare available tools and publish or contribute reusable work where permissions and licensing allow.

How you will work

Build with scientists and methodologists, combining software development with focused scientific review. Help reviewers make disagreement explicit and productive. Design the workflow so that maintaining the shared account requires less effort than reconstructing it repeatedly.

Lead on system evaluation against ordinary search and structured evidence review on the same questions and sources. Measure evidence quality, scope errors, review effort, correction and whether the work improves a real decision or hand-off. Preserve failed ideas, alternative interpretations and justified no-action decisions.

Reproducibility is central: a reader should be able to inspect the source and analysis behind a claim, understand its limitations and see what changed. Laboratory owners retain control over what is shared. Scientific and programme decisions remain with accountable people, supported by the software you build.

Influence how senior management embraces AI across the Institute to deliver its mission faster.

What you bring

• Strong programming ability and evidence of delivering useful software, with the ability to learn the languages and tools needed by the codebase.

• Sound engineering practice: Git with GitHub or equivalent, code review, tests, continuous integration, documentation, packaging, dependency management and maintainable releases.

• Experience modelling and integrating complex information, for example in scientific data systems, search, evidence synthesis, collaborative knowledge tools or research software.

• The ability to reason carefully about evidence, ambiguity, provenance and changing interpretations, and translate that reasoning into usable software.

• Practical experience applying and evaluating language models or related information-extraction/retrieval methods, with a clear account of their failure modes.

• Curiosity about how scientific communities establish knowledge and coordinate work, and a commitment to reproducibility, open collaboration and constructive disagreement.

• Enthusiasm and ability to influence how an organisation can drive change through technology adoption.

• The ability to work autonomously and lead: set priorities, make and justify technical and strategic decisions, bring others with you and take ownership of outcomes.

Experience with knowledge graphs, ontologies, discourse graphs, evidence synthesis, interface design, biomedical research or open-source scientific software would be valuable. Prior expertise in a particular graph database, a particular programming language or Alzheimer's research is not required. Scientific judgement will be developed with domain experts. We welcome varied professional and educational paths; no particular degree or PhD is required where equivalent capability can be demonstrated.