Head of AI – Agentic Security Systems

Lead the technical development and strategic direction of an agentic AI core for security operations.

Remote (South Africa), Industry: Cybersecurity, Working hours: Monday to Friday 8-5 pm (UK working hours).

Salary: Market Related

About Our Client

The client is an AI security team providing an agentic AI system that reasons across security signals, correlates incidents, and takes action with human oversight. The company is built on five years of MSSP operations and proprietary threat intelligence from over 100 clients. They are a revenue-generating entity that has recently closed a seed round.

The Role: Head of AI – Agentic Security Systems

The purpose of this role is to own the intelligence layer that differentiates the company and lead the strategic evolution of the AI from a prototype into a scalable, in-house intelligence engine. The role exists to lead the technical development, architecture, and roadmap of the agentic AI core while reporting directly to the CTO/CPO. The main focus areas involve migrating from cloud-dependent models to self-hosted open-source LLMs and ensuring the system is cost-efficient and production-ready.

Key Responsibilities

• Possess 7+ years in AI/ML engineering, with 3+ years on production LLM or agentic systems.

• Own the AI roadmap end-to-end, including planning, prioritisation, delivery, and reporting to leadership.

• Lead the strategic migration from AWS Bedrock to a self-hosted open-source LLM like Llama or Mistral.

• Extend and maintain the LangGraph-based multi-agent pipeline and the Strands cognitive layer.

• Provide technical direction and mentorship to existing data scientists and engineers.

• Implement cross-module incident correlation and behavioural anomaly detection.

• Manage LLM cost governance, token economics, and GPU infrastructure planning.

• Collaborate with backend engineers on FastAPI integration, AWS infrastructure, and GitLab CI/CD.

About You

• 7+ years in AI/ML engineering, with 3+ years on production LLM or agentic systems.

• Demonstrated ownership of an AI system end-to-end from architecture to production.

• Strong Python skills and experience with LangGraph, LangChain, or equivalent frameworks.

• Deep understanding of hybrid LLM architectures and when to use deterministic code vs. LLM inference.

• Experience with open-source LLM deployment, fine-tuning, RAG, and vector databases like Qdrant.

• Proficiency with AWS services, including Bedrock, Lambda, EC2, S3, and DynamoDB.

• Cybersecurity background and familiarity with Splunk, Cribl, or MITRE ATT&CK.

• Data analytics experience in anomaly detection, clustering, or user risk scoring.