Senior AI Infra Engineer: Self-Improving Agents

Neura Market; Greater London; Hybrid; GBP 110,000 - 150,000; Full time

Job summary

Together AI is seeking engineers to join the Infra Agent Systems team, building production AI agents for diagnosing hardware, incidents, and automation across one of the world’s largest GPU fleets. You’ll also help craft the platform powering these agents, including knowledge graphs and retrieval/orchestration capabilities.

You’ll work on two hard problems: making agents trustworthy for production, and building the knowledge, retrieval, and distributed systems that enable them to reason and act

Qualifications

• 5+ years of experience building production backend or distributed systems.
• Strong systems design skills and ownership of large-scale systems from design to production.
• Expertise in AI agent systems, knowledge graphs, or search/retrieval concepts.

Responsibilities

• Design and build production AI agent systems diagnosing, investigating, and remediating infrastructure issues across a large GPU fleet.
• Develop distributed services, orchestration, knowledge graphs, and retrieval systems powering infrastructure agents.
• Create fleet intelligence by fusing telemetry, state, and incidents to improve agent decisions.
• Integrate with observability, incident management, ticketing, and internal systems via robust APIs.
• Own services end-to-end from architecture to production operations and delivery.
• Improve agent performance through evaluation, retrieval improvements, and tooling.

Skills

Backend systems

Distributed systems

AI agents

Kubernetes

Go

TypeScript

Python

Rust

Tools

ArgoCD

Git

NATS

Kafka

Prometheus

Job description

Together AI is seeking engineers to join the Infra Agent Systems team, building production AI agents for diagnosing hardware, incidents, and automation across one of the world’s largest GPU fleets. You’ll also help craft the platform powering these agents, including knowledge graphs and retrieval/orchestration capabilities.

You’ll work on two hard problems: making agents trustworthy for production, and building the knowledge, retrieval, and distributed systems that enable them to reason and act