AI Engineer (Expert)

ATS Client; Pretoria; On-site; ZAR 1,310,186 - 2,292,826; Full time

Job summary

ATS Client is seeking a lead AI/ML architect to define agentic system architectures using Bedrock and agent frameworks, guiding model selection and performance trade-offs.

You will design containerized deployments with Docker/Kubernetes, build low-latency networking for model-service communication, and drive MLOps practices including CI/CD and versioning across teams.

Qualifications

• Proven experience in designing agentic architectures with Bedrock AgentCore.
• Orchestrate multi-step reasoning, tool invocation, and workflow automation for AI agents.
• Hands-on training and deployment of models with PyTorch and TensorFlow.
• Containerization using Docker and Kubernetes for scalable deployments.
• Networking for ML workloads including VPC design and low-latency communication.
• Experience with CI/CD for models and observability in ML systems.

Responsibilities

• Define and build agentic system architectures leveraging Amazon Bedrock and agent frameworks.
• Lead technical strategy for model selection, fine-tuning, and performance trade-offs.
• Design and implement containerized deployment standards using Docker and Kubernetes.
• Architect secure, low-latency networking for model-to-service communication.
• Perform systems-level performance engineering, including load testing and capacity planning.
• Establish MLOps practices, including CI/CD pipelines and model versioning.
• Integrate foundation models into enterprise workflows for complex use cases.
• Provide technical leadership and mentorship to engineers and stakeholders.

Skills

System Architecture Design

Multi-Step Reasoning

Networking for ML Workloads

MLOps Practices

Tools

Docker

Kubernetes

PyTorch

TensorFlow

Terraform

AWS Bedrock

AWS Lambda

Job description

Job Description

Define and build agentic system architectures leveraging Amazon Bedrock and agent frameworks.

Lead technical strategy for model selection, fine-tuning, and performance trade-offs.

Design and implement containerized deployment standards using Docker and Kubernetes.

Architect secure, low-latency networking for model-to-service communication.

Perform systems-level performance engineering, including load testing and capacity planning.

Establish MLOps practices, including CI/CD pipelines and model versioning.

Integrate foundation models into enterprise workflows for complex use cases.

Provide technical leadership and mentorship to engineers and stakeholders.

Requirements

Essential Skills

• System Architecture Design: Proven experience in designing and building agentic system architectures using frameworks like Amazon Bedrock AgentCore.
• Multi-Step Reasoning: Strong expertise in orchestrating multi-step reasoning, tool invocation, and workflow automation for AI agents.
• Model Training and Deployment: Deep hands-on knowledge of training and deploying models using PyTorch and TensorFlow.
• Containerization: Skills in Docker and Kubernetes for scalable and fault-tolerant ML/GenAI deployments.
• Networking for ML Workloads: Solid understanding of networking principles, including VPC design and low-latency communication patterns.
• MLOps Practices: Experience with CI/CD for models, model versioning, and observability in ML systems.

Advantageous Skills

• Cloud Services Experience: Prior experience with Amazon Bedrock and other cloud-managed foundation model services.
• Infrastructure as Code: Familiarity with tools like Terraform for reproducible cloud infrastructure.
• Serverless Architecture: Knowledge of serverless components (e.g., AWS Lambda) for event-driven workflows.
• Data Engineering: Experience in building reliable ETL/data pipelines for model training and feature stores.
• Observability Tools: Familiarity with observability stacks like Prometheus and Grafana for monitoring ML services.
• Enterprise Compliance: Understanding of compliance considerations in regulated industries (e.g., automotive, finance).