Senior Developer (AI Platforms – AI Journeys & Agentic Engineering)

Location: UK-Wide (Flexible, Hybrid) | Business Area: Development |

Reporting into the Chief Technical Officer, we’re looking for an experienced Senior Developer to build our AI discovery platform, deliver Connex AI journeys for our customers and lead the adoption of agentic AI across our development teams.

We're building ASK, an AI-powered customer discovery platform. A conversational assistant interviews a customer about their business, systems and pain points, costs each problem, ranks the opportunities into the customer’s own portal and matches them to AI “journeys” we deliver – virtual receptionist, appointment booking, cash collection, automated reporting, or bespoke consultancy. It is multi-tenant, runs on Azure in the UK and is live in UAT today, with Chess as its first customer. As Senior Developer you will own that platform and take it from a founder-built prototype to a product customers pay for.

This is a hands-on technical role. You will ship code most days: extending the ASK platform, building and integrating the Connex journeys customers switch on, and creating the tooling and working practices that make agentic AI a normal part of how our development teams deliver. You will work closely with the AI Lab, our AI consultants, Sales and Pre-Sales, and with customers during scoping, testing and go-live.

"a great place to work, a great place to be a customer"

Key Responsibilities:

• Building and owning the ASK platform end to end – Node/TypeScript on Azure Container Apps, PostgreSQL with row-level security per tenant, multiple identity providers per tenant, a pluggable LLM layer and usage billing – and taking it to production:

  • Pen-test remediation, production environment and sizing, custom hostname, outbound email and Microsoft 365 organisation import
  • Journey activation against Simply AI, Copilot agents and Connex, and integration with Nexus, our API layer for retrieving customer data during a live journey
  • Roadmap features: deep-dive question packs, connectors (Jira, Azure DevOps, ServiceNow, webhooks), the MCP endpoint, reporting and AI risk register modules

• Keeping the platform defensible: tenant isolation, audit trail, UK data residency, secrets in Key Vault and tests that run on every push.

• Designing, building and deploying agentic AI journeys on Connex for Chess customers – voice and chat automations such as virtual receptionist, appointment booking, cash collection calls and reporting – from discovery record through to live service.

• Building the integrations a journey needs to identify the caller, pull account status, book the slot and write back to the customer’s CRM, finance or ticketing system, working directly with REST APIs, webhooks and data payloads.

• Making journeys reusable – built once, with customer-specific data injected at runtime – and keeping the engine swappable so a journey can move between Connex, Simply AI and Copilot without a rebuild.

• Owning the technical side of our internal Connex deployment, including PCI scope, SSO, mobile app governance and pen-test findings.

• Working with our AI consultants, Sales and Pre-Sales to scope bespoke journeys and turn recurring requirements into repeatable catalogue items.

• Setting the working practices for AI-assisted and agentic development across our development teams – which tools we use (Claude Code, Copilot and others), what gets reviewed and how, and where the guardrails sit – in line with our agentic AI and code-generation risk treatment plans.

• Coaching engineers from occasional AI use to running agents as a normal part of their work, moving the team up our AI maturity model (Author → Editor → Director → Orchestrator).

• Building the internal tooling that makes this stick: MCP connectors to our systems, shared prompts and skills, CI checks and evaluation of new tools as they arrive.

• Measuring and reporting the effect on throughput, quality and adoption – plainly, including where it is not working.

• Ensuring code, integrations, journeys and decision logic are clearly documented and maintainable so nothing depends on a single person.