About the job

Full-time | Reports to Head of Pivotal Innovations Lab | Location: North / Hybrid

About Pivotal

Pivotal is on a mission to be UK's Trusted Adviser. Backed by Pollen Street Capital, we have built a family of specialist advice firms through an ambitious and continuing acquisition programme. Now we are building the technology that unites them: one AI-first tech stack - the Pulse Platform, with a one data layer, and an AI-first client experience for one of the largest advice markets in the country.

We're a people-first business where colleagues excel, grow and feel genuinely valued. Our entrepreneurial partners lead award-winning teams delivering exceptional advice with speed, precision and care. We're obsessed with the client experience, naturally curious, and never stop striving for better.

About the role

The Hub is Pivotal’s platform for making every mortgage case easier to process. At its heart is Jarvis, an AI assistant that draws on lender criteria, client and property information, firm rules, and years of historic case data to help advisers know exactly what to do next - and, in time, to process cases itself.

Jarvis is only as good as what it knows. The AI/Context Engineer owns that: every data source flowing into the Hub, the structure that makes it retrievable, and the evidence that answers are getting better. You will work with the Pivotal Innovations Lab’s innovators and engineers, and act as the bridge between Jarvis and the several hundred advisers and administrators who rely on it.

What you’ll do

• Build and run ingestion pipelines in production connecting lender, client, property, firm-rules and historic case data into the Hub.
• Grow what Jarvis knows: identify, assess and onboard new data sources, and maintain the source register recording each source’s provenance, classification and refresh cadence.
• Own how Jarvis's knowledge is organised: the structures, labels and retrieval set-up that determine whether Jarvis finds the right information - and the templates that shape how it uses it.
• Build and operate the evaluation harness - a firm-owned test set measuring answer accuracy and freshness - and publish the scores so improvement is visible.
• Establish governance (a year-one deliverable): define and embed the approval process for new sources, particularly anything touching client personal data.
• Enable the network (~15-20% of your time): own the documentation and “how Jarvis works” knowledge base, and train champions and team leads who cascade it to the wider firm.
• Fix what goes wrong: when Jarvis gives a wrong or out-of-date answer, trace it back to the cause - a missing source, stale data, or how the information was retrieved - and fix it at the root.

How success is measured

• Jarvis's answers get measurably better over time - and you have the test scores to prove it.
• The data feeding Jarvis grows in coverage and stays fresh, with every source documented.
• More advisers use Jarvis every week, and cases take less time to process.

How this role fits in

The engineering team owns the Hub’s application code and infrastructure. The Head of Innovations owns the product roadmap. Compliance owns sign-off on data use. You own everything between a raw source and the model’s context window.

What you’ll need

Essential:

• Production data engineering (Python, SQL, APIs, pipeline orchestration)
• Demonstrable hands-on LLM/RAG work in any setting (we build on AWS Bedrock and LangGraph) - side projects count, but be ready to show them
• The ability to explain technical systems in plain English to non-technical colleagues.
• You document exceptionally well - clear, complete and organised, so others can navigate your work.

Desirable:

• Mortgage or financial services domain knowledge
• Training experience
• Working knowledge of UK GDPR in a regulated environment
•