> Markdown version of [/jobs/ext/2650869-artificial-intelligence-ai-offensive-security](https://www.wearedevelopers.com/jobs/ext/2650869-artificial-intelligence-ai-offensive-security). Every page supports `.md` or `Accept: text/markdown`. Links point to the HTML versions so they work for humans too. Agent guide: [/agents.md](https://www.wearedevelopers.com/agents.md). --- # Artificial Intelligence (AI) Offensive Security... - **Company:** Citigroup, Inc. - **Location:** Tampa, FL, United States - **Experience:** Expert - **Salary:** $87,280.0 - $212,160.0 - **Contract:** Permanent contract - **Skills:** Application Programming Interfaces (APIs), Artificial Intelligence, Software System Penetration Testing, Application Portfolio Management, Automation of Tests, Burp Suite, Encodings, Cyber Security, Data Mining, Mobile Application Software, Nmap, Open Web Application Security, Red Team (Cyber Security), Software Vulnerability Management, Web Applications, Application Enhancement Tool, Large Language Models, Multi-Agent Systems, Mitre Att&ck, Generative AI, Information Technology, Build Process, Vulnerability Analysis - **Published:** August 8, 2026 - **Apply:** https://www.juju.com/job/00000000gm9bzs ## About the Role Security Testing Experience + Practical experience running penetration tests or vulnerability assessments against web applications, APIs, or mobile applications. Alternatively, Red Team experience running adversary simulations as an operator + Familiarity with common security testing tools: Burp Suite, nuclei, nmap, or similar + Working knowledge of OWASP Top 10 and at least some exposure to OWASP LLM Top 10 or MITRE ATLAS + Ability to triage findings and communicate risk clearly to technical and non-technical audience AI Literacy + Hands-on experience using LLM tools in a meaningful way, whether for security testing, security research, or your own workflows + Enough understanding of how LLMs work (context windows, tool use, RAG, agentic chaining) to reason about where they can go wrong + Familiarity with AI testing tools or frameworks, even from personal research or experimentation, is a real plus: Garak, PyRIT, PromptBench, Inspect AI, or equivalent General + Comfortable working on problems that don't have a playbook yet + Can write clearly: test plans, findings reports, risk summaries + Curious about the security implications of AI systems, not just the hype around them, + Bachelor's degree in Computer Science, Information Security, Engineering, or equivalent practical experience. + Master's degree preferred _The role sits within Citi's Offensive Security and Vulnerability Management (OSVM) organization, a team whose mission is to safeguard Citi's clients, partners, and business sectors by integrating an attacker's mindset across the technology lifecycle._ _We have multiple openings across various experience levels, from senior to principal, offering a dynamic environment for growth and impact. Location and compensation packages may be flexible and will be commensurate with experience and qualifications._, Agentic Design, AI Penetration Testing, Artificial Intelligence (AI), Cybersecurity, Large Language Models (LLMs), MITRE ATT&CK Framework, OWASP Top 10, Penetration Testing. ## Description The Chief Information Security Office (CISO) is home to deeply talented colleagues that work to ensure the safety of Citi's clients', our revenue, our employees and our proprietary data. We manage information security as one end-to end program - one with a clear mandate and accountability. Our mission is a program that is fully anchored to modern control and architectural frameworks, is fully aligned with the enterprise architecture of the firm and is deeply integrated into the businesses and functions. About the Role Citi's Offensive Security and Vulnerability Management (OSVM) team is hiring practitioners who can run security assessments against AI systems and use AI-powered tools to make security testing faster and more effective. The AI Offensive Security Analyst is an operational role on a team that runs penetration tests, vulnerability assessments, and Red Team adversarial evaluations across Citi's application portfolio. The AI dimension is twofold: you'll test AI systems as a target, and you'll use AI tools as part of how you work. We're looking for someone to help build the tools from scratch and who knows how to get the most out of them, knows their limits, and can operate in an environment where the right approach isn't always obvious yet. What Makes This Different The problems we're solving don't have playbooks yet. + **AI-driven vulnerability management.** When frontier models can generate thousands of findings in a weekend, the bottleneck shifts from discovery to triage, verification, and remediation. We're building the pipeline that makes this sustainable. + **Security architecture for the AI era.** Defining how a global bank deploys, monitors, and governs agentic AI - from standards and evaluation frameworks to production runtime monitoring. + **AI at scale, not in a lab.** Citi has deployed AI tools to 180,000+ employees, equipped 30,000 developers with AI coding assistants, and is rolling out agentic AI capabilities firm wide. Securing that footprint and the platforms that underpin them is the job. + **Securing AI agents that can behave like insider threats.** Frontier models can harvest credentials, escape sandboxes, and adapt when they detect monitoring. We're designing the containment architectures and runtime controls to operate them safely at enterprise scale. + **Real security engineering.** This isn't a cyber seat where you'll spend your time in administration . You will be expected to understand the code, the architecture, the threats, and find solutions. You'll have the mandate and the backing to build something meaningful. + **Cyber Security Operations** - Detection, triage, and response for a world where adversaries use AI to find and exploit vulnerabilities faster than traditional detection can keep up. Behavioral analytics for AI agents. Playbooks for AI-originated attack scenarios. What You'll Work On + **Testing AI applications and agents** . GenAI applications, agentic systems, and LLM-backed products require testing that goes beyond standard web app methodology. Prompt injection, indirect injection, retrieval poisoning, tool misuse, sensitive data extraction, unsafe delegation, these are real attack classes with real business impact. You'll design and execute test cases for them. + **Using AI to work better** . AI-assisted testing changes what's possible in a pen testing engagement: broader coverage, faster reconnaissance, automated validation of common vulnerability classes. You'll be expected to know how to use these capabilities effectively and to help the broader team adopt what works. + **Embedding security into development** . Working with development teams and security engineers to make sure AI application assessments happen at the right points in the build process., We use automated processing, including artificial intelligence, for our legitimate business interests (or our reasonable and appropriate business purposes) to identify and align the candidate's skills and abilities with a specific job opening. Additionally, if you so choose, or consent, we can match your skills and abilities to other suitable roles at Citi. Importantly, all our hiring processes and decisions, including determining your suitability for a role, are conducted, checked, and decided by individuals. Our automated processing and AI do not involve relying on automatic or autonomous decision-making. Please refer to any Jurisdictional Considerations, with specific provisions for your country (where relevant) for further details. Illinois residents - AI Notice and Right (https://tbcdn.talentbrew.com/company/287/cms/v3/docs/policies/Illinois\_Career\_Supplement\_a11y.pdf) ## Related Videos - [How to Cause (or Prevent) a Massive Data Breach- Secure Coding and IDOR](https://www.wearedevelopers.com/videos/39-how-to-cause-or-prevent-a-massive-data-breach-secure-coding-and-idor) - [A Brief History of Data Storage](https://www.wearedevelopers.com/videos/974-a-brief-history-of-data-storage) - [Spot, Squash, Secure: Fighting Security Bugs with GitHub Copilot](https://www.wearedevelopers.com/videos/100052-spot-squash-secure-fighting-security-bugs-with-github-copilot) - [JSON and Beyond](https://www.wearedevelopers.com/videos/968-json-and-beyond) - [The New AI Security Stack: Observe, Detect, Protect](https://www.wearedevelopers.com/videos/100302-the-new-ai-security-stack-observe-detect-protect) - [Bitcoin transactions- embedding and using data for applications built on Bitcoin SV](https://www.wearedevelopers.com/videos/152-bitcoin-transactions-embedding-and-using-data-for-applications-built-on-bitcoin-sv) ## Related Articles - [Dev Digest 134 - Where pixels sing?](https://www.wearedevelopers.com/magazine/477-dev-digest-134-where-pixels-sing) - [Got AI ideas but no money? Here are 10 free ways to level up your AI skills with Google Cloud](https://www.wearedevelopers.com/magazine/600-got-ai-ideas-but-no-money-here-are-10-free-ways-to-level-up-your-ai-skills-with-google-cloud) - [Dev Digest 137 - AI'm not sure about this](https://www.wearedevelopers.com/magazine/485-dev-digest-137-ai-m-not-sure-about-this) - [Navigating the AI Shift](https://www.wearedevelopers.com/magazine/629-navigating-the-ai-shift) - [MLOps And AI Driven Development](https://www.wearedevelopers.com/magazine/82-mlops-and-ai-driven-development) - [Dev Digest 132 - Binging WADFlix?](https://www.wearedevelopers.com/magazine/473-dev-digest-132-binging-wadflix)