DontApply.Club
Login Register

AI Security Engineer

Waftx — Nepal · Posted ~2 hours ago

Skills

Cybersecurity Application security Penetration testing Red teaming LLM security Prompt injection testing Jailbreak testing Security test case design YAML JSON Docker Git Command line APIs Security findings documentation LLMs Python

🔓 Log in to save this job, tailor your resume & track your apply process — 7 days free, no card needed.

Log in to add to target list

Summary ✨ AI‑Generated

Work as an AI Security Engineer focused on adversarial testing of large language models. You will create prompt-injection and jailbreak test cases, execute them against AI systems, validate outcomes, and document findings. The role combines offensive security skills with emerging LLM security techniques and requires comfort with Docker, Git, command-line tools, APIs, and structured data formats. Python and prior AI security research are valuable additions.

Highlights

Hands-on opportunity to research and test emerging AI security threats, develop structured attack scenarios, validate results, and contribute to practical LLM security research.

Description

Role Description Create prompt injection test cases, run them against AI models, verify the outcomes, and document successful results. Qualifications Experience in Cyber Security, Application Security, Penetration Testing, or Red Teaming.Strong understanding of LLM security, including prompt injection, jailbreaks, and AI attack techniques.Ability to design and test prompt injection attacks against AI models.Experience writing structured test cases (YAML, JSON, or similar formats).Comfortable using Docker, Git, command line tools, and APIs.Familiarity with Python scripting is a plus.Experience validating security test results and documenting findings.Knowledge of one or more business domains such as e-commerce, finance, healthcare, enterprise SaaS, or HR is preferred.Background in AI Security, LLM Security Research, Offensive Security, Bug Bounty, or Ethical Hacking is highly preferred.