Senior Software Engineer - AI Platform Engineer

Design, build, and operate Go services running on Kubernetes that form the backbone of our agent infrastructure

Own reliability for production systems - define SLOs, write runbooks, and be accountable when things go wrong

Instrument everything: structured logging, distributed tracing (OpenTelemetry), and metrics that surface both system health and the behaviour of the AI workflows running on top

Contribute to Kubernetes-native patterns: operators, CRDs, workload isolation, and resource management for dynamic agent workloads

Build and maintain CI/CD pipelines that support rapid, safe iteration - using AI tooling as a natural part of that workflow

Use AI development tooling effectively - generating, reviewing, and steering code with the same rigour you’d apply to any PR

Decompose and delegate work to AI agents where appropriate, and know when not to - catching drift, validating output, and maintaining ownership of outcomes

Review AI-generated code with an understanding of its specific failure modes: plausible-looking but subtly wrong logic, missing edge cases, and over-confident implementations

Contribute to the team’s shared judgement on when new models, frameworks, or tooling change the architecture assumptions we’re building on

Partner with the broader AI Foundation team to establish platform patterns that other teams adopt

Participate in architecture reviews and contribute to build vs buy decisions

Mentor more junior engineers and raise the team’s bar for production readiness and operational discipline

You'll work closely with ML engineers and agentic systems engineers on the team, providing the platform foundation - reliability, observability, deployment infrastructure - that lets their work reach production safely.

This is a genuine collaboration: you'll bring rigour to how their systems are built and operated, and they'll bring you along on the AI domain specifics - model behaviour, context design, evaluation approaches - as the work demands.

You won't be expected to arrive knowing that space deeply; you will be expected to engage with it seriously.

5+ years of professional software engineering experience, with meaningful time operating at senior level

Strong Go - idiomatic, well-tested, production-shipped (if you meet all the other requirements, but have a different language - please apply)

Solid Kubernetes operations experience: you’ve debugged real production incidents, understand the scheduler, and know when a CRD is the wrong answer

Hands-on observability experience - OpenTelemetry, Prometheus, distributed tracing - and a genuine conviction that unobservable systems are untrustworthy

Cloud-native background on AWS or GCP: IAM, managed Kubernetes, infrastructure-as-code

Comfortable working in an AI-augmented development environment - directing tools, critically evaluating their output, and maintaining engineering rigour throughout

Demonstrated ability to own production systems end-to-end: design, ship, monitor, iterate

Familiarity with Tekton, Jenkins, Argo Workflows, or similar pipeline infrastructure

Exposure to how LLM-based systems are structured - enough to reason about them as infrastructure dependencies, not black boxes

Experience contributing to platform or shared infrastructure used by multiple teams

Shipping production code to the AI Foundation platform with minimal ramp-up

Owning at least one service end-to-end: SLOs, alerts, and runbooks included

Identifying and addressing a meaningful observability gap in an existing workflow

Delivering a materially improved platform capability that the team relies on

Contributing patterns or tooling that other engineers adopt

Being the person teammates reach for on K8s and observability questions

Leading a significant platform initiative from design through to production reliability

Influencing how the team approaches build vs buy vs integrate decisions

Raising the bar on how the team uses AI tooling in its own engineering practice

Platform engineering here isn’t maintenance work. The systems are genuinely hard - dynamic workloads, heterogeneous signal sources, reliability requirements that don’t bend because the underlying models are probabilistic.

You’ll also be engineering in the way most teams will be working in a few years: AI tooling as a normal part of the workflow, with engineers who know how to get the most out of it without losing ownership of what ships.