Machine Learning Operations (MLOps) Engineer (GCP)

Anyone can say they've "deployed ML models on GCP."

Very few can tell you the QPS, the p99 latency, and what went wrong at 2am.

TMC Middle East is hiring a Consultant MLOps Engineer (GCP) on behalf of a leading organization in Saudi Arabia.

This is not a pipeline-builder role.

This is a production ownership role.

You will operationalize, monitor, and scale ML systems that carry real traffic, real SLOs, and real consequences when they fail.

If you've lived that — read on.

What you'll actually own:

→ End-to-end MLOps architecture on GCP.

You design it, you build it, you run it.

→ CI/CD automation for ML — from Git commit to canary deployment on Vertex AI Endpoints.

Every step automated, every step auditable.

→ Production monitoring — drift detection, alerting, automated retraining triggers.

You know before the business does when something is wrong.

→ Incident response.

When a model degrades in production, you're the person with the answer — root cause, rollback, post-mortem, prevention.

What we need to see — non-negotiable:

✦ You deployed a production ML system at scale.

Tell us the QPS, the latency SLO, and one failure mode you personally resolved.

✦ You wrote GCP code within the last 12–18 months.

Active.

Recent.

Not theoretical.

✦ You know Vertex AI from the inside — pipelines, model registry, endpoints, monitoring.

War stories preferred.

✦ You have a production incident story.

Real root cause.

Real fix.

Real post-mortem action.

✦ You can whiteboard your last GCP ML architecture cold — no notes, no prep.

✦ 5+ years ML Engineering / MLOps / DevOps. 3+ years GCP mandatory.

GCP Professional Cloud DevOps Engineer cert is a strong plus.

What we don't want:

✗ "I've worked with Vertex AI" — with no endpoint, no monitoring config, no numbers

✗ POC builders who've never carried a production SLO

✗ Anyone whose GCP muscle is 2+ years stale

Stack: GCP · Vertex AI · Kubeflow · Cloud Build · Docker · Python · Terraform

One question before you apply:

What's the most painful Vertex AI limitation you've hit in production — and how did you work around it?

If you have an answer ready, we want to hear from you.

Please share your cv f to Abishake Kosman : [email protected]