Multi-Agent AI Platform Engineer

Dex; Greater London; On-site; GBP 100,000 - 150,000; Full time

Job summary

Dex connects you with partner companies building autonomous AI platforms. This role focuses on crafting production-grade, multi-agent infrastructure for a fast-growing property-tech acquisition engine.

You will design memory, tool-calling, orchestration, and evaluation layers enabling agents to handle live workflows across landlords, tenants, and service providers, with safeguards and deterministic backends.

Qualifications

• Shipping LLM-powered or agentic systems in production, not just demos.
• Strong backend foundation in TypeScript/Node.js or Python with TS-first capability.
• Demonstrated architectural judgement on where to deploy probabilistic LLM agents vs deterministic logic vs human review.
• Active daily use of AI coding agents to decompose tasks, write code, and accelerate development.

Responsibilities

• Architect the core platform for multi-agent orchestration, context management, tool integration, and state persistence across operational workflows.
• Build automated evaluation harnesses and observability tooling to benchmark agent reliability and catch regressions before systems go live.
• Standardise custom operational processes into modular, reusable agentic components for onboarding of new businesses in days.
• Define the architectural boundary between autonomous agent execution, deterministic backend code, and human intervention safeguards.
• Leverage agentic software engineering practices and coding agents as a core part of daily development cadence.

Skills

LLM-powered systems

Agentic systems

Backend engineering

TypeScript/Node.js

Python experience

Architectural judgement

AI coding agents

Tools

Node.js

Python

Job description

Dex connects you with partner companies building autonomous AI platforms. This role focuses on crafting production-grade, multi-agent infrastructure for a fast-growing property-tech acquisition engine.

You will design memory, tool-calling, orchestration, and evaluation layers enabling agents to handle live workflows across landlords, tenants, and service providers, with safeguards and deterministic backends.