AI Security Engineer
Exnessgloballimited
·- Location
- Cyprus · CY Office
- Department
- Security
- Education
- Master
Description
At Exness, we are not just a leading trading broker—we’ve reimagined what it takes to be a leader. With 2,000+ people across 13 countries, we combine scale, care, and real tech to make trading better for over a million clients globally.
Recognised globally as a Best Place to Work, we’re a people-first company where long-term wins always matter more. As part of our team, you will shape the future of fintech with real technology, care, and purpose.
Why this role matters
As an AI Security Engineer, you will help build the security foundation that allows Exness to use LLMs, MCP integrations and autonomous agents safely at scale. You will focus on making sure AI systems can access data, tools and critical services without creating risks.
A big part of the role is turning security policies and threat models into real, testable technical controls. You will work on the centralized AI gateway, build guardrails and policy enforcement for LLM and agent workloads, secure MCP integrations and tool access, design sandboxing and isolation controls for agent execution, create a human-in-the-loop service approach with risk scoring for critical actions, set up Shadow AI detection and use AI to automate security processes. You will also audit and harden the systems that make up the AI platform, from the gateway to on-prem model serving.
We are looking for a hands-on senior engineer who is comfortable working across software development, infrastructure security, identity and LLM engineering, including fine-tuning and serving models. You should be able to take a security problem, design a practical solution, ship it to production and measure whether the control actually works.
The goal is to ensure that every AI interaction, whether started by an employee, a service or an autonomous agent, goes through controls that are enforced outside the model, observable, measurable and independently verifiable, without slowing down adoption
The role is based in our Limassol office, Cyprus. In case of relocation, we offer full relocation support for you and your family to make your move smooth and worry-free.
What you'll actually do
- Make the AI gateway the single controlled path to SaaS and on-prem models, and keep the platform behind it secure. This includes routing by data classification and security telemetry.
- Own the controls that govern what AI models receive, return, and can do through tools. Developing guardrails against prompt injection and data leakage, with fine-tuned classifiers where needed, and running the security review and onboarding of MCP servers on a least-privilege basis.
- Enable AI agents to do real work without exceeding the access of the users they act for. Design and implement the approach where agents run in isolation, with no more access than their user, and a human-in-the-loop service approach that sends critical actions for human approval based on risk scoring.
- Find AI use outside approved channels and bring it under control. This includes setting up the process and tooling to detect cases such as personal AI accounts, unregistered MCP servers and working with SOC and IT to block, replace, or onboard each one.
- Scale AI security works with AI itself, so that reviews and monitoring keep pace with adoption. This includes automating security review triage, assessments of MCP servers and AI tools, and analysis of AI security events on top of LLMs, agents, and fine-tuned on-prem models.
Who we’re looking for
- Bachelor's or Master's degree in Computer Science, Software Engineering, Information Security, Applied Mathematics / Machine Learning or a related field, or equivalent practical experience.
- Certifications are not mandatory. Relevant examples: CKS (Certified Kubernetes Security Specialist), OSCP / OSWE, cloud security certifications. Courses or certifications in ML/LLM engineering or AI security are a plus.
- 5+ years in security engineering, platform / infrastructure engineering or backend development with a strong security focus.
- 2+ years of hands-on work with LLM / GenAI systems in production: building LLM-based services or agents, serving or fine-tuning models, or securing AI platforms.
- A track record of designing, shipping and operating production services and security tooling, not only reviews and audits.
- Experience in fintech, trading or other regulated, high-load environments is an advantage.
- LLM engineering and automation: solid practical understanding of how LLMs work and fail (context, sampling, tool calling, structured outputs, embeddings and RAG, agent loops, non-determinism); proven experience building LLM-powered automations or agents used in production, including evaluation (test sets, metrics, calibrated LLM-as-judge) and cost / latency control.
- Fine-tuning, deployment and optimization of models: hands-on fine-tuning of open-weight models with PyTorch and the Hugging Face stack (Transformers, PEFT, TRL), including dataset preparation and evaluation, e.g. for classification and detection tasks; serving and optimizing inference with vLLM (quantization, batching and KV-cache tuning, parallelism, GPU memory sizing, latency and throughput benchmarking) on Kubernetes.
- AI gateway: hands-on experience with LiteLLM, or equivalent depth with another AI gateway.
- Software engineering: strong Python or at least one systems language (Go, Rust, C/C++) for production services and APIs with tests, CI/CD and containerization; building and operating services with observability (metrics, logs, traces) and SLOs.
- Infrastructure security: Linux and Kubernetes security (RBAC, admission policies, Pod Security, NetworkPolicy, runtime isolation; network security (egress control, proxies, DNS, TLS / mTLS); identity (OAuth 2.1 / OIDC, JWT, token exchange, workload identity); secrets management; policy-as-code (OPA / Rego); hardening, vulnerability and patch management; IaC (Terraform, Helm, GitOps).
- AI security: practical knowledge of AI-specific attacks and controls (direct and indirect prompt injection, jailbreaks, tool poisoning, data exfiltration through agents, excessive agency, model and AI supply-chain risks); working knowledge of MCP and its authorization model, the OWASP Top 10 for LLM and for Agentic Applications, and MITRE ATLAS.
- Builder and ownership mindset: takes a capability from idea to production and operations, and owns the outcome rather than the task.
- Platform thinking: solves classes of problems with reusable controls and secure defaults instead of one-off fixes.
- Pragmatic, risk-based judgement: balances security, usability and delivery speed; avoids over-gating that pushes people towards Shadow AI.
- Autonomy: plans own work in a small team and raises blockers early.
What we offer along the way
- Competitive salary and annual performance bonus
- Full relocation support for you and your family — flights, housing, visas, and legal assistance included
- Top-tier health insurance with full family coverage — medical, dental, vision, mental health — plus life insurance for peace of mind
- Unlimited learning opportunities: external courses, English lessons, career and leadership development
- Education allowance covering school and kindergarten fees
- 21 working days of annual leave, plus public holidays and fully paid sick, maternity, and paternity leave
- Employee appreciation program: branded gifts, celebration budgets for weddings, newborns, and milestones
- “Get to know Team” trips — meet colleagues across our global hubs, along with company-wide offsites that raise the bar
- Employee share scheme — grow with us
- Branded MINI Cooper Countryman company car and private parking
- Free in-house sports clubs, Sanctum Club gym access, and jet skis
- Access to a Corporate doctor
- Exclusive discount program with cafes, gyms, and local services
- Expat tax perks: up to 50% income tax exemption
- Support with the naturalisation process for relocated employees
At Exness, we know that changing jobs - and changing countries - is a big step. That’s why relocation with Exness is different. We make it smooth, supported, and truly life-changing.
What your journey looks like
- Intro call with Recruiter (up to 45 minutes)
- Short online English test (for non-native speakers)
- 1st Technical interview (60 minutes)
- 2nd Technical interview (60 minutes)
- Behavioural interview (60 minutes)
What it's like here
Curious about what working at Exness really looks like? Follow us on Instagram and LinkedIn.
We share the real Exness experience - our people, ideas, moments, and everything in between.
Sounds like you? Apply.
Please note: We occasionally amend or withdraw Exness jobs and reserve the right to do so at any time, including prior to the advertised closing date. Before applying, you are advised to read our data protection policy. This policy describes the processing that may be associated with your personal data and informs you that your personal data may be transferred to Exness/Exness Group companies around the world. Exness Group and its approved recruitment consultants will never ask you for a fee to process or consider your application for a career with Exness. Anyone who demands such a fee is not an authorized Exness representative and you are strongly advised to refuse any such demand.
At Exness, we're an equal opportunity employer where every individual is valued. No matter your race, color, religion, sex, national origin, sexual orientation, gender identity or disability, we welcome you. As an international fintech company, we embrace the richness of our diverse team, respecting each individual and promoting gender equality for all genders in our workforce.