Customer Cluster Engineer

The role

You will be embedded with three to five reserved-capacity accounts at a time, owning their training and inference performance end-to-end.

You're paired with a single account manager who handles commercial; you handle the technical relationship from kickoff through renewal.

You carry the customer-engineering on-call rotation alongside runtime and cluster SRE — about one week per six.

What you'll do

• Own three to five reserved-capacity accounts as their named customer engineer. Most are AI labs or AI-native scale-ups running 256–2048-GPU jobs.
• Profile distributed training jobs: NCCL collectives, gradient overlap, checkpoint cost, MFU. Make and defend specific recommendations.
• Tune kernels and configurations alongside the runtime team — your changes go through the same review process and ship in the same release train.
• Run the technical pre-sales for your accounts' expansions and renewals: pilot design, scope, success criteria, post-mortem.
• Lead the post-mortem on any P1 affecting your accounts; escalate root-cause work into the runtime, cluster, or platform teams.
• Carry the customer-engineering on-call rotation alongside runtime and cluster SRE — about one week per six.

The bar

• Five-plus years of distributed-systems or ML-systems engineering — production experience with 64+ GPU jobs is required.
• Strong PyTorch / FSDP / Megatron-LM debugging skills. You can read a wandb run and tell us where time is going.
• Working knowledge of CUDA and at least one of Triton or CUTLASS. You don't need to write them daily; you need to read them.
• Strong communicator. You'll write four post-mortems a quarter and present at customer all-hands.
• Comfort with executive-level conversations on technical scope, timelines, and trade-offs.

Bonus

• Nice to have, not required:
  • Prior frontier-lab or hyperscaler training-team experience.
  • Experience running benchmarks at customer sites (MLPerf or similar).
  • Open-source contributions to PyTorch, FSDP, NCCL, or the major training frameworks.
  • Comfort with Slack-based customer relationships and high-tempo asynchronous communication.

Compensation

In writing, like everything else
We publish bands. We meet them. The number you see on the offer is the same number your future peers got at the same level.
We do not negotiate; we level.
$220,000 – $320,000 USD (US, IC4 / IC5).

Meaningful early-stage equity, refreshed on tenure milestones.
No revenue-tied compensation. Customer engineers are not compensated on the size or renewal of the accounts they own; they are compensated on technical impact.

How to apply

• One email is enough
  Send a short note to careers@iframe.ai with the role title in the subject line. Include your CV or LinkedIn, one or two links to work you're proud of, and a sentence on why this role specifically.
• Hiring managers reply within five business days, regardless of outcome.
• 01ApplicationA hiring manager reads every email.
• 02Manager call30–45 minutes. Scope, role, mutual fit. We share the comp band on this call.
• 03Technical loop3–4 sessions on the same day. Real problems, no homework, no whiteboard riddles.
• 04OfferSame-week offer at the published band for your level. Start dates are flexible.

Also open

• Runtime Inference Acceleration Engineer Ship the next 2× on the open-source model catalog. Triton, CUDA, ROCm. You will publish what you ship.
• Research Lab Distributed Training Researcher Lead a paper / quarter on multi-thousand-GPU pre-training. Co-appointment with a partner university available.
• Cluster & SRE Cluster Site Reliability Engineer Bring up B300 racks, drive InfiniBand fabric to spec, run capacity planning across seven regions. Pager included.

One last thing

If this role isn't quite right but you'd be fit at iframe.ai, write anyway.
Senior engineers and researchers can apply outside the listed roles. The bar is the same. The reply window is the same.