MLOps Engineer Resume Example
Pipeline layout: a train-deploy-monitor node header, platform metrics and a model-platform table for MLOps and ML platform engineers.
Use this example Template: MLOps Engineer
Omar Haddad
MLOps Engineer — ML Platforms, LLM Serving & Model Monitoring
omar.haddad@example.com+971 55 555 0264Dubai, United Arab Emirateslinkedin.com/in/omarhaddad-mlops
10Years engineering
40Models live
2 daysModel release
−60%Model incidents
Summary
MLOps engineer with 10 years in software and data platforms, the last 4 running production ML for a regional super-app. I build the path from notebook to production: feature store, CI/CD for models, GPU serving on Kubernetes and drift monitoring. My platform serves 40 models and 2 fine-tuned LLMs at 18k requests per second; model release time fell from 3 weeks to 2 days and GPU cost per inference by 46%.
Experience
Senior MLOps Engineer, ML Platform
Careem · Dubai
Mar 2021 – Present
40Models in production
18k rpsPeak inference
−46%GPU cost / inference
- Built the model release pipeline (MLflow registry, GitHub Actions, Argo CD) with automated evaluation gates; release time fell from 3 weeks to 2 days
- Moved LLM and vision serving to vLLM and Triton on autoscaled GPU node pools, cutting GPU cost per inference by 46% at p95 latency under 120 ms
- Rolled out a Feast feature store shared by pricing, ETA and fraud teams, removing 30% of duplicated feature pipelines
- Added drift, data-quality and LLM evaluation monitoring (Evidently, Prometheus, Grafana) with on-call runbooks; model incidents fell by 60%
Data Engineer, Analytics Platform
Emirates NBD · Dubai
Jun 2018 – Feb 2021
12 TBDaily ingestion
- Built Spark and Airflow pipelines feeding credit-risk and churn models; first bank team to deploy models as containerised APIs
- Introduced data contracts and Great Expectations tests, cutting broken downstream reports by 70%
Backend Engineer
Fetchr · Dubai
Jul 2016 – May 2018
- Built Python and Go services for route optimisation and address geocoding used by 1,500 couriers
Platform at a glance
| Layer | Stack | Result |
|---|---|---|
| Training | Kubeflow, Ray | Spot GPUs, −35% cost |
| Registry & CI/CD | MLflow, Argo CD | 3 weeks → 2 days |
| Serving | vLLM, Triton | p95 < 120 ms |
| Features | Feast, Redis | −30% duplicate pipelines |
| Monitoring | Evidently, Grafana | −60% incidents |
Summary example
MLOps engineer with [N] years in software and data platforms, the last [N] running production ML for a regional super-app. I build the path from notebook to production: feature store, CI/CD for models, GPU serving on Kubernetes and drift monitoring. My platform serves [N] models and [N] fine-tuned LLMs at [N] requests per second; model release time fell from [N] weeks to [N] days and GPU cost per inference by [X%].
Example bullet points for a MLOps Engineer
- Built the model release pipeline (MLflow registry, GitHub Actions, Argo CD) with automated evaluation gates; release time fell from [N] weeks to [N] days
- Moved LLM and vision serving to vLLM and Triton on autoscaled GPU node pools, cutting GPU cost per inference by [X%] at p95 latency under [N] ms
- Rolled out a Feast feature store shared by pricing, ETA and fraud teams, removing [X%] of duplicated feature pipelines
- Added drift, data-quality and LLM evaluation monitoring (Evidently, Prometheus, Grafana) with on-call runbooks; model incidents fell by [X%]
- Built Spark and Airflow pipelines feeding credit-risk and churn models; first bank team to deploy models as containerised APIs
- Introduced data contracts and Great Expectations tests, cutting broken downstream reports by [X%]
- Built Python and Go services for route optimisation and address geocoding used by [N] couriers
Replace the [brackets] with your own numbers and facts. Never claim results you didn't achieve.
Skills to include
- Model CI/CD
- LLM serving & evaluation
- Feature stores
- Drift monitoring
- Kubernetes
- MLflow
- vLLM / Triton
- Terraform
- Python / Go
Tips for a MLOps Engineer resume
- Put the hard requirements recruiters screen for in the top third: Model CI/CD, LLM serving & evaluation, Feature stores.
- List current credentials with the issuing body, e.g. Certified Kubernetes Administrator (CKA), AWS Certified Machine Learning — Specialty.
- Give every recent role at least one measured result — for example models in production.
- Match the job title in your headline to the posting where it is honest to do so.
- Run the built-in resume checker and paste the job description into the keyword match before you download.
Related: How to write a resume · Action verbs · All templates