Data & MLOps Engineer

Description

Ubuntu Tech Solutions is a proudly South African firm and a BBBEE Level 1 contributor. We act as a strategic partner to our clients, helping them solve complex challenges across Data and AI, Technology, and Risk Our core mission is to empower organizations by:

• Modernizing digital platforms

• Unlocking strategic value from their data

• Navigating complex regulatory and financial risk.

This role ensures the integrity of AWS-based data platforms and AI/ML inference infrastructure. You will support high-availability data lakes, ensuring that feature pipelines do not serve stale data to live models. You are responsible for platform health, capacity management, query optimization, and the operational requirements of compute-intensive AI workloads.

Key Job Responsibilities

• Manage AI/ML training and inference infrastructure on AWS, including Sage

Maker and EC2 GPU instances.

• Support model deployment pipelines and monitor model serving endpoints for drift or degradation.

• Operate feature stores and model registries.

• Perform data pipeline scheduling, monitoring, and issue resolution for services like AWS Glue and Redshift.

• Execute performance tuning, query optimization, and storage lifecycle management.

• Conduct regular restore testing and backup/recovery operations.

• Support Generative AI infrastructure, specifically Amazon Bedrock and associated LLM operations.

• Implement resource scheduling and cost governance for compute-intensive AI and ML workloads.

• Execute data platform patching and upgrade management to maintain vendor supported versions.

Basic Qualifications

• 5+ years of experience with Redshift, AWS Glue, S3 data lakes, RDS, and DynamoDB.

• Hands-on experience with Amazon Sage

Maker and model management tooling like MLflow.

• Experience with MLOps practices and GPU compute management.

Preferred Qualifications

• Experience with environment reproducibility and dependency management for AI/ML workflows. Familiarity with data governance and cataloging tooling.

Johannesburg, Sandton (Onsite/Hybrid)