Senior MLOps Engineer

Job description / Role

About the Institute of Foundation Models (IFM)

The Institute of Foundation Models is a dedicated research lab for building, understanding, deploying, and risk-managing large-scale AI systems. We drive innovation in foundation models and their operationalization, empowering research, education, and industry adoption through scalable infrastructure and real-world applications.

As part of our engineering team, you will operate at the intersection of machine learning and systems design — building the cloud, orchestration, and deployment layers that power the next generation of intelligent applications at MBZUAI. You’ll work alongside world-class AI researchers and engineers to productionize LLMs, voice models, and multimodal systems at scale.

The Role

As a senior MLOps engineer, you will design, build, and maintain robust machine learning infrastructure across training, inference, and deployment pipelines. You will take ownership of the model lifecycle — from data ingestion to real-time serving — and ensure our LLM and speech models are deployed efficiently, securely, and reproducibly in Kubernetes-based environments.

This position requires deep hands-on experience with Kubernetes (EKS), Helm, AWS cloud infrastructure, and modern MLOps toolchains (e.g., vLLM, SGLang, OpenWebUI, Weights & Biases, MLflow). Familiarity with speech and voice AI frameworks like ElevenLabs, Whisper, and RVC is also valuable.

Key Responsibilities

• Design and manage scalable machine learning infrastructure on AWS using EKS, EC2, RDS, S3, and IAM-based access control.
• Build and maintain Kubernetes deployments for LLM and TTS inference using Helm, ArgoCD, and Prometheus/Grafana monitoring.
• Implement and optimize model serving pipelines using vLLM, SGLang, TensorRT, or similar frameworks for high-throughput inference.
• Develop CI/CD and MLOps automation for data versioning, model validation, and deployment (GitHub Actions, Jenkins, or AWS CodePipeline).
• Integrate OpenWebUI, Gradio, or similar user interfaces for user-facing model demos and internal evaluation tools.
• Collaborate with machine learning researchers to productize models — including TTS (e.g., ElevenLabs API), ASR (Whisper), and LLM-based chat systems.
• Ensure observability, cost optimization, and reliability of cloud resources across multiple environments.
• Contribute to internal tools for dataset curation, model monitoring, and retraining pipelines.
• Maintain infrastructure-as-code using Terraform and Helm charts for reproducibility and governance.
• Support real-time multimodal workloads (voice, text, vision) across inference clusters.

Academic Qualifications

• 4+ years of experience in MLOps, DevOps, or cloud infrastructure engineering for machine learning systems.
• Strong proficiency in Kubernetes, Helm, and container orchestration.
• Experience deploying machine learning models via vLLM, SGLang, TensorRT, or Ray Serve.
• Proficiency with AWS services (EKS, EC2, S3, RDS, CloudWatch, IAM).
• Solid experience with Python, Docker, Git, and CI/CD pipelines.
• Strong understanding of model lifecycle management, data pipelines, and observability tools (Grafana, Prometheus, Loki).
• Excellent collaboration skills with machine learning researchers and software engineers.

Professional Experience – Preferred

• Extensive experience with vLLM, Kubernetes, ElevenLabs, Whisper, Gradio/OpenWebUI, or custom TTS/ASR model hosting.
• Familiarity with multi-GPU scheduling, NCCL optimization, and HPC cluster integration.
• Knowledge of security, cost management, and network policy in multi-tenant Kubernetes clusters and cloudflare systems.
• Prior work in LLM deployment, fine-tuning pipelines, or foundation model research.
• Exposure to data governance and responsible AI operations in research or enterprise settings.

We may use artificial intelligence (AI) tools to support parts of the hiring process, such as reviewing applications, analyzing resumes, or assessing responses and identifying potential inconsistencies or verification signals in application materials based on available information. These tools assist our recruitment team but do not replace human judgment. Final hiring decisions are ultimately made by humans.