AI Engineer - Cyber Programs

Coca-Cola; Georgia; On-site; USD 150,000 - 190,000; Full time

Job summary

The Coca-Cola Company is seeking an AI Engineer for Cyber Programs in a senior advisory-capable role. You will serve as the company’s dedicated adversarial tester for AI systems, designing and executing engagements to uncover vulnerabilities in AI models, applications, and infrastructure.

Expect to develop methodologies, tools, and reporting to harden AI deployments enterprise-wide. This senior individual-contributor position partners with governance, development, and external specialists to

Qualifications

• 8–12 years of cybersecurity experience with 3–5 years in AI/ML security or applied AI research with security focus.
• Hands-on adversarial testing of AI systems including LLMs, agents, and AI-enabled apps.
• Strong understanding of AI architectures, training pipelines, and inference APIs.

Responsibilities

• Plan and execute adversarial testing engagements against Coca-Cola’s AI models, applications, and infrastructure.
• Develop AI red team methodologies, playbooks, and tooling aligned with industry standards.
• Assess AI security controls, governance processes, and incident readiness for AI-specific events.
• Coordinate with external consultants, bug bounty programs, and AI security researchers.
• Produce clear adversarial testing reports with evidence, risk ratings, and remediation guidance.

Skills

AI/ML security

Adversarial testing

Red team operations

Threat modeling

Security governance

Vulnerability research

Cross-functional collaboration

Strong communication

Experience with LLMs

Tools

Garak

PyRIT

Job description

Job Description Summary

Overview

The AI Engineer - Cyber Programs is The Coca-Cola Company's dedicated adversarial testing specialist for AI systems. This role systematically probes the Company's AI models, AI-enabled applications, and AI infrastructure for vulnerabilities, ensuring that AI deployments across the enterprise are resilient against adversarial manipulation, misuse, and emerging attack techniques. The focus is on AI-specific adversarial testing - not traditional network or application penetration testing - covering risks such as prompt injection, jailbreaking, data poisoning, model exfiltration, training data leakage, and abuse of agentic AI capabilities.

This is a senior individual-contributor role that combines deep expertise in AI/ML systems with offensive security tradecraft applied to the AI domain. The role designs and executes AI cyber-program engagements, develops testing methodologies and tooling, and delivers clear, actionable findings that enable the Company to harden its AI systems before and after deployment. The role partners with AI governance, development teams, and external specialists (including consultants and bug bounty programs) to ensure comprehensive adversarial coverage of the Company's AI portfolio.

Key Responsibilities

AI Adversarial Testing & Red Teaming

• Plan and execute adversarial testing engagements against the Company's AI models, LLM-based applications, agentic AI systems, and AI infrastructure.
• Test for AI-specific vulnerabilities including prompt injection, jailbreaking, output manipulation, data poisoning, model inversion, training data extraction, membership inference, and adversarial evasion.
• Evaluate the security of AI system architectures, including model hosting, API exposure, data pipelines, fine-tuning workflows, and agent tool-use chains.
• Assess risks introduced by third-party AI components, vendor-embedded AI, and AI-enabled SaaS products.

Methodology & Tooling Development

• Develop and maintain AI red team methodologies, playbooks, and testing frameworks aligned with NIST AI RMF, MITRE ATLAS, OWASP Top 10 for LLMs, and emerging industry standards.
• Build and maintain custom tooling, scripts, and automation for AI adversarial testing, including prompt fuzzing, model probing, and automated vulnerability scanning of AI systems.
• Evaluate and integrate open-source and commercial AI security testing tools (such as Garak, PyRIT , or similar) into the testing workflow.
• Stay current with the latest AI adversarial research, published attacks, and defensive techniques.

AI Security Controls Assessment

• Assess the effectiveness of AI security controls, including input/output filtering, content moderation, guardrails, access controls, logging, and monitoring.
• Evaluate human oversight mechanisms, model governance processes, and incident response readiness for AI-specific security events.
• Test the resilience of AI systems under adversarial conditions, including coordinated multi-step attacks and advanced jailbreak chains.
• Recommend proportionate controls and mitigations based on testing results, risk severity, and the Company's AI risk appetite.

External Partnerships & Coordination

• Coordinate with external consultants, specialized AI security firms, and bug bounty programs for AI-specific adversarial assessments.
• Engage with the AI security research community, industry working groups, and standards bodies to stay current and contribute the Company's perspective.
• Partner with AI governance, development teams, and business units to ensure testing priorities align with the Company's AI deployment roadmap.

Reporting & Knowledge Building

• Produce clear, actionable adversarial testing reports with findings, evidence, risk ratings, and remediation guidance for technical and executive audiences.
• Maintain a knowledge base of AI vulnerabilities, attack patterns, and testing results to inform future assessments and track remediation progress.
• Brief Cyber Defense leadership and the CISO on AI threat trends, adversarial testing results, and the Company's AI security posture.

Qualifications

• Minimum 8-12 years of progressive cybersecurity experience, with at least 3-5 years focused on AI/ML security, adversarial testing of AI systems, or applied AI research with a security focus.
• Demonstrated hands-on experience conducting adversarial testing of AI systems, including large language models, machine learning models, agentic AI, or AI-enabled applications.
• Strong technical understanding of AI/ML systems, including model architectures, training pipelines, inference APIs, fine-tuning, retrieval-augmented generation (RAG), and agentic frameworks.
• Working knowledge of AI-specific attack techniques, including prompt injection, jailbreaking, data poisoning, model inversion, training data extraction, and adversarial evasion.
• Familiarity with AI security frameworks and guidance, including NIST AI RMF, MITRE ATLAS, OWASP Top 10 for LLMs, and ISO/IEC 42001.
• Experi