AI Platform Engineer
Own Stem, the core Marrow runtime that turns agent graphs into compliant, auditable product journeys. Compiler design meets infrastructure meets regulated AI.
- Team
- Engineering
- Location
- London
- Type
- Full-time
- Working pattern
- Hybrid - we're flexible, but do think we do our best work, quickest, together
- Salary
- £75k - £100k
- Equity
- Meaningful equity
About this role
We're a small, senior founding team and this is one of our most important engineering hires. You'll own the core Marrow platform, called Stem - the runtime that turns agent graphs into compliant, auditable product journeys. Think of it as compiler design meets infrastructure meets regulated AI. Hard? Yes! Fun? Yes!
The problem is genuinely novel. No one has built a production-grade compliance layer for AI insurance agents before - and, as such, you won't be following a playbook - you'll be writing one. That means meaningful authority over technical architecture from day one, and real accountability for the outcome.
You'll play a key role in
- Designing and owning the Marrow platform runtime - how agents are compiled, configured, and deployed at scale
- Building the configuration layer that lets carriers control and iterate on product journeys without writing code
- Owning the data model behind the systems and tooling - building the engine that allows us to rapidly meet customer needs, however they develop
- Defining the security model: PII handling, data residency, model access controls
- Building self-improvement loops - performance tracking, automated evaluation across deployments
- Creating the observability and audit trail that makes a regulated insurer comfortable running AI at scale
- Continuously improving Marrow's agentic engineering position through automated validation, sophisticated CI pipelines and agentic review
We'd love to chat if
- You've operated at staff or principal engineer level building serious backend infrastructure
- You have strong opinions about distributed systems, API design, and how to build for auditability
- You have prior experience serving robust systems to a large user base
- You're comfortable with LLMs and agent frameworks - you don't need to have shipped a production agent platform (almost no one has), but you understand the constraints well
- You think naturally about security - not as a checklist, but as part of how you design things
- You want to be the person who defines what this company is technically capable of, not the person who executes someone else's architecture
- No insurance background needed - we'll give you what matters
Not ticking every box? That's fine. If you're genuinely excited by the problem and have most of what we're describing, we'd rather hear from you than not.
What's in it for you
- Founding engineer title and meaningful equity from day one
- A small team that moves fast and trusts you to make calls
- London-based hybrid working - we're flexible, but do think we do our best work, quickest, together
The interview process
We keep it short. Three stages:
- 0130-minute intro call with Tim (CEO) to talk about the role and your background
- 021 hour technical session with Adam (CTO), featuring systems design and a light touch project walkthrough
- 0330-minute culture and goals fit with Adam and Tim
We'll move as fast as you need us to.
Sound like you?
Applications are handled on Wellfound. We reply to everyone, usually within a few days.