AI Security Analyst

Location: Sofia, Bulgaria; full-time onsite role

Company

A technology organization focused on cross-functional collaboration across countries and cultures to develop secure, resilient, and trustworthy systems.

What you will do

• Assess security risks across AI/ML models, training data, data pipelines, APIs, and deployment environments.
• Identify and mitigate AI-specific threats, including data poisoning, model inversion, prompt injection, model theft, and adversarial attacks.
• Design and implement access controls, monitoring, anomaly detection, and vulnerability monitoring for AI systems and related tools.
• Conduct threat modeling and risk assessments for AI-enabled applications across design, training, testing, deployment, and retirement.
• Contribute to AI security standards, policies, guardrails, and responsible AI practices.
• Support incident investigations and develop playbooks for AI-related security incidents.

Requirements

• Bachelor’s degree in Cybersecurity, Computer Science, Engineering, or equivalent experience.
• Strong experience in cybersecurity, application security, cloud security, or a related field.
• Hands-on experience with AI/ML platforms, frameworks, or workflows, including model training, inference, or MLOps.
• Familiarity with AI threat models and attack techniques such as adversarial machine learning, prompt injection, and model leakage.
• Experience securing AI in cloud-native environments, including containers, CI/CD, and managed ML services.
• At least one year of experience with ticketing systems, SIEM, DLP, and vulnerability management tools or programs.

Nice to have

• Knowledge of AI governance, model risk management, or responsible AI principles.
• Relevant cybersecurity certifications.
• Experience with ServiceNow, Cylance, Falcon, Carbon Black, Proofpoint, Rapid7, or Tanium.

Culture & Benefits

• Collaborative, enriching work environment focused on continuous improvement.
• Cross-functional collaboration across countries and cultures.
• Workplace based on mutual respect, open communication, and shared execution.