Staff AI Engineer - Platform AI

Hyperlayer; Greater London; On-site; GBP 120,000 - 180,000; Full time

Benefits offered by this job

Performance-based bonus

25 days annual leave + birthday + ten­

Critical illness and life insurance

AXA PPP Health Insurance

Cycle-to-work

Referral bonus

Workplace Nursery Scheme

Job summary

Hyperlayer is building a cutting-edge AI-driven banking platform and seeks a Principal AI Engineer to lead AI infrastructure and the agent layer. You’ll shape end-to-end AI architecture, governance, and scalability across a multi-function system.

The role combines hands-on design with strategic decisions in a security-conscious, regulated environment. You’ll lead design reviews with peers, drive production readiness, and collaborate across data, platform, and product teams to ensure reliable,

Qualifications

• 8+ years of experience in AI/ML or software engineering, including designing and shipping agentic systems into production, not just prototypes.
• Experience architecting agent governance for high-stakes domains: guardrails, autonomy levels, kill switches, and human-in-the-loop approval
• Experience with agent runtimes - sandboxing and isolation, tool invocation (e.g. Model Context Protocol), multi-agent orchestration, and reasoning-loop control
• A track record of building or operating an evaluation and analytics harness for agent behaviour: trace capture, hallucination detection, and regression testing across releases
• Comfort with cloud platforms (AWS mainly, GCP is a plus) and container tech (Docker, Kubernetes)
• Experience designing systems that combine deterministic, rules-based processing with agent-based reasoning in the same architecture
• Solid Python and SQL skills, and hands‑on experience with LLM or agent frameworks (LangChain or similar)
• Understanding of identity and authorisation models extended to non‑human/agent principals - delegated authority, composite human‑plus-agent identity
• A data‑driven mindset - comfortable defining the metrics that tell you whether an agent is actually working
• Strong communication and collaboration skills
• Curiosity and a love for experimentation, always exploring what’s next in how AI gets embedded in complex, regulated systems

Responsibilities

• Design the agentic control plane for a new platform — how agents get authored, deployed, governed, run, and observed across a complex, multi-function system
• Define how agents are declared and scoped: registry, authority boundaries, autonomy levels, and tool bindings, compiled into the same versioned release artefacts as everything else
• Own the evaluation and analytics layer for agent behaviour: trace capture, hallucination detection, containment measurement, cost accounting, and regression tracking across releases
• Participate in system level design and architecture decisions with your peers
• Make sure our AI systems scale, stay reliable, remain secure, and meet financial regulation objectives
• Work closely with the platform’s core architecture and engineering leads to keep the agentic layer and the rest of the system cohesive, not two products bolted together
• Shape the AI architecture for a platform still being built - your early decisions will define how agents behave across the whole system for years to come.

Skills

Experience in AI/ML

Agent governance

Agent runtimes

Evaluation analytics harness

Cloud platforms AWS/GCP

Docker/Kubernetes

Deterministic + agent systems

Python & SQL

LLM/agent frameworks

Identity & authorization models

Data-driven mindset

Communication & collaboration

Curiosity & experimentation

Tools

LangChain

Model Context Protocol

Job description

About Hyperlayer

Hyperlayer is the intelligent banking platform built for the agentic economy: the financial control layer that sits above banks' existing cores, enabling next-generation Smart Accounts, without ripping anything out. From a single integration, we give financial institutions a secure rules and orchestration engine that lets them design, launch and scale hyper-personalised products in weeks while preserving stability, compliance and control. And there's nothing like us anywhere in the world.

About Hyperlayer

Hyperlayer is the intelligent banking platform built for the agentic economy: the financial control layer that sits above banks' existing cores, enabling next-generation Smart Accounts, without ripping anything out. From a single integration, we give financial institutions a secure rules and orchestration engine that lets them design, launch and scale hyper-personalised products in weeks while preserving stability, compliance and control. And there's nothing like us anywhere in the world. Hyperlayer works with Tier 1 and Tier 2 FIs across the UK, US, EMEA and Hong Kong, with an award-winning UK consumer app (HyperJar) sitting in the group. We're creating the fintech category that every bank in the world needs: programmable, governed, agent-ready accounts. Your job is to make sure our audiences know that.

We’re Growing Our AI Team

Hyperlayer is on a mission to reinvent how people and businesses move money, and we need a Principal AI Engineer with big ideas to help us make it happen. This isn’t just about writing code; it’s about building the AI infrastructure for the future of payments.

You’ll provide the AI leadership for bringing a complex, multi-function platform to life — designing the agent layer that sits alongside a much larger system of data, rules, and reporting capability, from initial architecture through to launch. If you love shaping how AI gets built into a large, many-part system end-to-end, you’ll fit right in.

What You’ll Do

• Design the agentic control plane for a new platform — how agents get authored, deployed, governed, run, and observed across a complex, multi-function system
• Define how agents are declared and scoped: registry, authority boundaries, autonomy levels, and tool bindings, compiled into the same versioned release artefacts as everything else
• Own the evaluation and analytics layer for agent behaviour: trace capture, hallucination detection, containment measurement, cost accounting, and regression tracking across releases
• Participate in system level design and architecture decisions with your peers
• Make sure our AI systems scale, stay reliable, remain secure, and meet financial regulation objectives
• Work closely with the platform’s core architecture and engineering leads to keep the agentic layer and the rest of the system cohesive, not two products bolted together
• Shape the AI architecture for a platform still being built - your early decisions will define how agents behave across the whole system for years to come.

Qualifications

• 8+ years’ experience in AI/ML or software engineering, including designing and shipping agentic systems into production, not just prototypes
• Experience architecting agent governance for high-stakes domains: guardrails, autonomy levels, kill switches, and human-in-the-loop approval
• Experience with agent runtimes - sandboxing and isolation, tool invocation (e.g. Model Context Protocol), multi-agent orchestration, and reasoning-loop control
• A track record of building or operating an evaluation and analytics harness for agent behaviour: trace capture, hallucination detection, and regression testing across releases
• Comfort with cloud platforms (AWS mainly, GCP is a plus) and container tech (Docker, Kubernetes)
• Experience designing systems that combine deterministic, rules-based processing with agent-based reasoning in the same architecture
• Solid Python and SQL skills, and hands‑on experience with LLM or agent frameworks (LangChain or similar)
• Understanding of identity and authorisation models extended to non‑human/agent principals - delegated authority, composite human‑plus-agent identity
• A data‑driven mindset - comfortable defining the metrics that tell you whether an agent is actually working
• Strong communication and collaboration skills
• Curiosity and a love for experimentation, always exploring what’s next in how AI gets embedded in complex, regulated systems

Bonus Points For

• Experience building or defining agent-to-agent or tool-integration protocols (Model Context Protocol and similar) at platform scale
• Knowledge of evaluation frameworks and fine‑tuning techniques.
• Familiarity with graduated autonomy models and agent-as-principal identity patterns
• Experience with prompt-injection and jailbreak defence, and other agent-specific security controls
• Experience with persona‑or‑scenario‑based simulation for testing agent behaviour before release
• Familiarity with hybrid deterministic/agentic execution models, where rules-based and LLM-based reasoning share the same runtime
• Experience with token, spend, or step‑budget governance for autonomous agents operating at scale
• Understanding of security and privacy in AI workflows (encryption, anonymization, compliance).
• Experience standing up an AI or agent capability from a blank sheet - architecture, governance and delivery - rather than extending an existing one

(You don’t need every skill. If you’re passionate about building the future with AI, we want to hear from you.)

Working at Hyperlayer

At Hyperlayer we are proud of our in-office culture. It makes a real difference to the working atmosphere and maximises collaboration in and across teams. This is critical to how we work and learn. Also, we think it makes for a great workplace!

We value our team and are proud to offer some great benefits:

• A performance‑based employee bonus scheme.
• 25 days annual leave plus bank holidays, a day off on your birthday, and an additional day per year of tenure to a maximum of 30.
• Critical illness and life insurance cover.
• AXA PPP Health Insurance, including gym offers.
• Cycle-to-work scheme.
• Referral bonus scheme.
• Workplace Nursery Scheme

Equity, diversity, and inclusion

Hyperlayer’s mission is to make a positive impact on people’s lives, making it easier for everyone to manage their money.

Hyperlayer’s commitment to equity, diversity, and inclusion is a key part of this; it helps us better serve our customers, employees, and business partners from every background.

We are constantly learning and striving to do better, and we expect all staff to actively take part in this process.

We will not be offering UK work visa sponsorship for these roles. Candidates will need an existing non-sponsored right to work in the UK.

Data Privacy

As part of our recruitment process, Hyperlayer Limited collects and processes personal data relating to job applicants.

We are committed to being transparent about how we collect and use this data and to meeting our data protection obligations. To read more click here.