Machine Learning Engineer

Penta Consulting; Gauteng; On-site; ZAR 900,000 - 1,500,000; Full time

Job summary

Penta Consulting in South Africa seeks an experienced ML Operations Engineer to build and maintain end-to-end MLOps and CI/CD pipelines for deploying models and infrastructure-as-code.

You will monitor performance, latency, data quality and system stability across enterprise workloads including LLMs and RAG pipelines, while enforcing model risk guidelines.

Collaborate with Data Scientists, IT and Data Engineering to operationalise models effectively and ensure compliance.

Qualifications

• Hands-on production ML lifecycle experience.
• Proficiency with PyTorch or TensorFlow in enterprise settings.
• Experience with monitoring and MLOps tooling.

Responsibilities

• Oversee production ML models and systems for stability and latency.
• Ensure model risk governance and regulatory compliance.
• Collaborate with Data Scientists, IT and Engineering to operationalise models.

Skills

Python scripting

PyTorch

TensorFlow

Containerisation

AWS

Azure

MLOps

Kubeflow

Tools

MLflow

Kubeflow

Docker

Job description

• Pipeline Management: Build and maintain end-to-end MLOps/AIOps and CI/CD pipelines for deploying models and infrastructure-as-code.
• Production Oversight: Monitor model performance, latency, data quality and system stability across enterprise workloads, including LLMs, RAG pipelines and predictive models.
• Governance & Compliance: Ensure that deployed AI and ML systems adhere to Model Risk Management guidelines, regulatory expectations and responsible AI frameworks.
• Collaboration: Partner with Data Scientists, IT and Data Engineering teams to operationalise models effectively and smoothly.

Responsibilities

• Production Oversight
• Governance & Compliance

Qualifications

• Technical Stack: Strong proficiency in Python scripting, major ML frameworks such as PyTorch or TensorFlow, containerisation, and cloud platforms such as AWS or Azure.
• Experience: Hands-on experience managing production machine-learning lifecycles, working with monitoring and MLOps tools such as MLflow or Kubeflow, and supporting enterprise data architectures.