For infrastructure & platform engineers
Our client is an AI-native cloud company. AI teams use its platform to run and scale AI workloads on large GPU clusters. The fleet is growing fast, and the work is automating it. That's where you come in.
HQ San Francisco · Engineering hub Amsterdam · Offices in London, Bangalore, Pune, Singapore
What you'd do
Your job: make every step of this lifecycle faster and more automated.
New data centers and GPU tranches imported into inventory and provisioning systems
Automated hardware validation and burn-in
Ansible/AWX, custom images, Kubernetes on bare metal or KubeVirt
Clusters go to internal teams or external customers
Triage failed nodes, RMA, retest, then turn the fix into code and agents
The fleet is growing fast, and the work is automating it. The goal is to make bring-up dramatically faster and fully automated.
Everyone works on automation. SREs here design processes and platforms, so software skills matter as much as operations.
The team builds agents that diagnose and fix infrastructure issues. Agents write code, humans review.
Almost all bare metal: latest-generation NVIDIA GPUs on InfiniBand and RoCE fabrics.
Linux, Ansible/AWX, Terraform, Python and Go (increasingly for AI tooling, incident collection and agents). Kubernetes controllers and custom images; bare-metal provisioning and automated hardware testing; KubeVirt, Kubernetes networking.
Compute platform keeps GPU capacity available for customers. Storage runs Ceph, filesystems and high-cardinality metrics. Every team wants the same profile: an SRE who codes like a platform engineer.
How the work is organised
Engineers own their outcomes and work closely as a team.
Three SRE teams: Fleet bring-up, Compute platform and Virtualization platform. They overlap, and engineers move between them as needed.
About half in Europe (Amsterdam/UK hours), half in the US, a couple in Singapore. Offices in Amsterdam and SF; most people, managers included, work remotely. No Pacific hours required.
Weekly Kanban cycles and a daily standup. No sprint planning or estimates. Slack for communication, and anyone can write an RFC.
Follow-the-sun: 12 hours on, 12 off, split between the US and Europe. Your turn comes about every six to nine weeks. You shadow first, after 1–2 months, and hand off when your window ends.
Claude Code, Cursor, Copilot, Open Code, Pi and internal agent tooling. Managers use AI daily. You stay the architect: AI speeds up the work, you own the decisions.
Through real work: tickets from week one, runbooks, and pairing with whoever joined just before you. A top-spec MacBook Pro ships to you; running a local Linux VM is fine.
Ownership and understanding problems independently, bringing problems together with possible solutions, clear communication at both depth and surface level, and understanding the whole system, not just your part.
After joining
“I also love the team and our manager. How they treat each other and work together. It's been a pleasure so far.”
“Two words: awesome & professional.”
“I'm super-happy with both how the team's been incredibly supportive and with myself being this adaptable. I'd have laughed if a year ago someone had told me I'd be doing SRE work a year later.”
“Remote and async like that requires you to be more intentional. They are very professional at what they do. They take this real seriously… It's just a great place to be in.”
“Everyone there, they are really good engineers, really good collaboration within the team. I really like the team. This type of work is not that easy to get nowadays… it's going to be a really good record for everyone's CV.”
“They shipped a MacBook Pro to my home… I surely wasn't expecting an M4 Max. That's not something most startups would do.”
Open through Lemon · as of Oct 13, 2026
You can be from anywhere, but you must live and work from one of these countries. Working elsewhere, even briefly, needs your manager’s and IT’s approval first. The list can change.
Interview Q&A
Scale. The company runs a large, fast-growing GPU fleet, while the team grows much more slowly, so the work is automating it. The goal is to make bring-up dramatically faster and fully automated.
Hands-on tickets first, to learn the system and the node lifecycle: triage, returning nodes to clusters, provider tickets. After a few weeks, the focus shifts to automating those problems. Experienced engineers move into agentic work as early as possible.
A blend. Everyone gets to work on automation. SREs here design processes and platforms, so strong software skills matter as much as operations.
Every SRE here is expected to think like an architect, and the role is now titled AI Infrastructure Engineer. Recent, heavy software experience is what it takes to build agentic solutions to the scale problem.
No. Strong Linux fundamentals matter most, and the team learns as it goes. One Lemon engineer moved from software development into SRE work here and found the team very supportive while he learned.
Follow-the-sun: 12 hours on, 12 off, with the US and Europe each covering their half. Your turn comes about every six to nine weeks. New joiners shadow first, usually after one to two months, and you aren't expected to work past your window.
Very. Managers use and advocate for AI every day. People use Cursor, Claude Code and Open Code. You stay the architect, with controlled, deterministic agent workflows.
About half the team works European hours (Amsterdam and UK) and half is in the US, with a couple of people in Singapore. Many people, including managers, work remotely. You don't have to work Pacific hours.
US, Canada, Mexico, EU countries, UK, India, Australia, Singapore, Norway, Switzerland, Iceland and Monaco. You must live and work from an approved country, and working from anywhere else, even briefly, needs your manager's and IT's approval first.
Ownership. You understand problems without needing your manager and bring them with possible solutions. Senior engineers propose how to stop incidents from recurring. Impact, clear communication and leading the solution count more than speed.
Pick it from the list and apply in a few minutes.
Talk to us
Apply to Lemon.io, get vetted once, and we’ll match you with projects that suit you best.
© Lemon.io, 2026 Privacy Policy Terms of Use