Dołącz do zespołu bezpieczeństwa AI.
CyberPolak buduje zespół, który zabezpiecza systemy sztucznej inteligencji: modele, aplikacje oparte na LLM, agentów AI i łańcuch dostaw oprogramowania. Rekrutujemy na 12 stanowisk, we współpracy B2B. Opisy stanowisk są po angielsku.
- 12stanowisk
- 1formularz, CV w PDF
- B2Bmodel współpracy
Pracujemy w modelu B2B. Zakres obowiązków, lokalizację (zdalnie lub hybrydowo) i widełki wynagrodzenia ustalimy na pierwszej rozmowie. Twoje CV trafia wyłącznie do zespołu rekrutacyjnego CyberPolak. Pytania o rekrutację: kariera@cyberpolak.edu.pl.
Złóż CVOpen positions
Kliknij stanowisko, aby zobaczyć pełny opis. Przycisk „Apply” wybierze je w formularzu.
01AI/ML Security EngineerEngineering
You will own the security of the AI systems CyberPolak builds and runs for its clients, from LLM features and retrieval pipelines to agent integrations. The goal is simple: AI ships fast, and it ships safe.
What you will do
- Threat-model LLM applications, RAG pipelines and agent tools for prompt injection, data leakage and tool abuse.
- Design and deliver controls: guardrails, output filtering, secrets handling and identity for non-human workloads.
- Add security checks to CI/CD and MLOps pipelines.
- Lead the response when an AI system behaves in an unexpected or unsafe way.
What we are looking for
- Solid security engineering background, ideally in cloud (AWS, Azure or GCP).
- Working knowledge of the OWASP Top 10 for LLM Applications.
- Python skills good enough to build and automate controls.
- Clear writing for engineers and decision makers.
02AI Application Security EngineerEngineering
A specialist track inside AI security engineering. You will secure AI-enabled software and code written with AI assistants: APIs, prompts, vector stores, model integrations and the tools agents can call.
What you will do
- Run architecture reviews and threat models for AI-enabled products.
- Test AI features and the code around them, including code generated by AI assistants.
- Write secure-coding guidance that developers actually follow.
- Build security gates into the delivery pipeline.
What we are looking for
- Hands-on application security experience: SAST, DAST, SCA and API security.
- Understanding of LLM-specific risks on top of classic AppSec.
- Ability to read and review code in at least one modern language.
- A pragmatic approach to developer experience.
03AI Red Team EngineerOffensive
You will attack CyberPolak's AI systems before real adversaries do. You will break language models, retrieval systems and agents, then turn every finding into something engineers can reproduce and fix.
What you will do
- Build multi-step prompt-injection and jailbreak chains against LLM apps and agents.
- Run adversarial machine learning attacks: evasion, poisoning, model extraction and inference.
- Test agent permissions and tool use for privilege escalation paths.
- Automate testing with open-source AI red-teaming frameworks.
What we are looking for
- Real penetration testing or red team experience, including cloud and Kubernetes.
- Curiosity about how language models fail.
- Python for building attack tooling.
- Reports with clear impact and clear fixes.
04AI Governance and Security Risk LeadGovernance
You will turn AI risk into controls the business can live with. You will own the rules for how AI is chosen, approved and used across the organisation, and you will explain them to leadership and auditors.
What you will do
- Run AI risk assessments covering security, privacy and fairness.
- Maintain the inventory of AI use cases and the list of approved tools.
- Write policies and controls mapped to the EU AI Act, ISO/IEC 42001 and the NIST AI RMF.
- Assess vendors and third-party AI before adoption.
What we are looking for
- Experience in GRC, risk or audit, ideally in a regulated sector.
- Strong working knowledge of the EU AI Act and AI management standards.
- Confidence presenting evidence to executives and regulators.
- Enough technical depth to challenge engineering teams.
05AI Security ArchitectArchitecture
You will define how AI is built securely across the organisation and sign off designs before they reach production. Reference architectures, trust boundaries and the final "not yet" are yours.
What you will do
- Own reference architectures for models, retrieval, inference and agents.
- Set trust boundaries, identity models, agent permissions and data protection.
- Map controls to recognised frameworks such as the NIST AI RMF and MITRE ATLAS.
- Review vendor and cloud AI platform designs.
What we are looking for
- Several years in security architecture, especially cloud and identity.
- Understanding of LLM, RAG and agent system design.
- Experience writing standards that teams adopt.
- The judgement and confidence to say no when a design is not safe.
06AI Threat Intelligence AnalystIntelligence
You will track how adversaries use AI and how AI systems are attacked, and turn that knowledge into action for the security operations team.
What you will do
- Collect and analyse intrusion data, OSINT and underground sources.
- Use LLM tooling to speed up analysis, and verify every output.
- Write detection rules (YARA, Sigma) from your findings.
- Brief the SOC and leadership on what has changed and what to hunt for.
What we are looking for
- Hands-on cyber threat intelligence experience and MITRE ATT&CK mapping.
- Python and SQL for collection and enrichment.
- A healthy scepticism of automated output.
- Concise, decision-ready writing.
07AI SOC OrchestratorOperations
You will run a security operations centre where AI does the first pass and you own the verdict. You will design the automation, supervise it and stop it when it is wrong.
What you will do
- Supervise AI-assisted alert enrichment, triage and investigation.
- Design and test the playbooks and agent workflows behind them.
- Write detections as code (Sigma, KQL, SPL).
- Escalate consequential decisions to a human, always.
What we are looking for
- Hands-on SOC experience with a SIEM such as Splunk, Microsoft Sentinel or Google SecOps.
- Experience with SOAR or workflow automation.
- Practical use of LLM APIs for triage.
- Calm judgement under pressure.
08AI Incident Response OrchestratorResponse
Attackers now move in minutes. You will lead incident response at that speed, using AI and automation for the heavy lifting while keeping evidence and decisions under human control.
What you will do
- Own incidents end to end: scoping, containment, eradication and reporting.
- Use AI to correlate evidence across cloud, endpoint and identity telemetry.
- Build response playbooks and automated workflows.
- Protect evidence integrity and build timelines that hold up.
What we are looking for
- Incident response or DFIR experience, including cloud environments.
- Scripting in Python or Go.
- Experience with forensic tooling and evidence handling.
- Clear communication during a crisis.
09AI Security ResearcherResearch
You will find the attack paths nobody has documented yet: in models, in agents and in the frameworks around them. Then you will make sure they stay fixed.
What you will do
- Discover and reproduce vulnerabilities in AI models, agents and their tooling.
- Build evaluation harnesses and regression tests for every finding.
- Research model extraction, distillation and agent manipulation techniques.
- Turn research into detections, guardrails and published guidance.
What we are looking for
- A track record in vulnerability research: CVEs, papers, tools or advisories.
- Reverse engineering, fuzzing or exploit development skills.
- Python with PyTorch or TensorFlow.
- Patience and rigour.
10Agentic AI Security EngineerEngineering
You will secure AI systems that take actions, not only answer questions. Agents with tools, memory and access to company systems need tight, testable limits.
What you will do
- Design least-privilege identity, authorisation and runtime controls for agents.
- Control tool and MCP permissions, memory, sandboxes and approval steps.
- Find and fix confused-deputy, unsafe tool use and goal-hijack paths.
- Monitor agent activity and build tests that keep agents within scope.
What we are looking for
- Understanding of agent loops, tool calling and the Model Context Protocol.
- Experience with identity, tokens and authorisation design.
- Python or TypeScript, and daily use of AI coding tools.
- An attacker's mindset.
11AI Supply Chain Security EngineerEngineering
You will make sure nothing enters the AI pipeline without proof of where it came from: models, datasets, packages and the build systems that ship them.
What you will do
- Verify the provenance of models, datasets and dependencies.
- Sign and scan artefacts and maintain SBOMs and ML-BOMs.
- Harden model registries, MLOps pipelines and CI/CD runners.
- Catch malicious or hallucinated packages before they ship.
What we are looking for
- DevSecOps or software supply chain security experience.
- Familiarity with Sigstore and SLSA.
- Hands-on CI/CD hardening.
- Attention to detail.
12Post-Quantum Cryptography Migration SpecialistCryptography
An adjacent role outside AI security. You will prepare CyberPolak's cryptography for the post-quantum era, following the new NIST standards (FIPS 203, 204 and 205).
What you will do
- Build an inventory of cryptography in systems, certificates and keys.
- Assess post-quantum readiness of HSMs, certificate authorities and key management.
- Plan and run hybrid classical and post-quantum migrations.
- Advise teams on crypto-agility.
What we are looking for
- Experience with PKI, HSM or key management.
- Understanding of the NIST post-quantum standards.
- Methodical project delivery.
- Ability to explain cryptography to non-specialists.
Złóż CV
Wypełnienie zajmie dwie minuty. Odpowiemy na każdą aplikację e-mailem.
- Przeglądamy Twoje CV i odpowiadamy e-mailem.
- Krótka rozmowa z zespołem CyberPolak. Omawiamy rolę, model B2B, warunki i widełki.
- Spotkanie techniczne i decyzja. Jeśli pasujemy do siebie - zaczynamy współpracę.
Nie wpisuj w CV danych, których nie potrzebujemy: numeru PESEL, zdjęcia, informacji o zdrowiu, wyznaniu czy poglądach.
Twoje dane w rekrutacji
Tak przetwarzamy Twoje dane osobowe, gdy wysyłasz nam CV (art. 13 RODO).
Kto jest administratorem
CYBERPOLAK Adam Jakóbaszek, ul. Rajska 5/70, 02-654 Warszawa, NIP 1131859545. W sprawach danych: privacy@cyberpolak.pl.
Po co i na jakiej podstawie
Żeby przeprowadzić rekrutację, na którą aplikujesz: na podstawie Twojej zgody i działań przed zawarciem umowy (art. 6 ust. 1 lit. a i b RODO). Do przyszłych rekrutacji tylko za osobną zgodą. Do ewentualnej obrony przed roszczeniami: nasz uzasadniony interes (art. 6 ust. 1 lit. f RODO).
Komu przekazujemy dane
Rekrutujemy do własnego zespołu, więc nie przekazujemy Twoich danych innym firmom w celach rekrutacyjnych. Korzystamy wyłącznie z firm, które obsługują nasz serwer i pocztę. Dane nie opuszczają Europejskiego Obszaru Gospodarczego.
Jak długo je przechowujemy
Do końca rekrutacji, najdłużej 6 miesięcy. Jeśli zgodzisz się na przyszłe rekrutacje: 12 miesięcy. Potem usuwamy aplikację razem z CV automatycznie.
Jakie masz prawa
Dostęp do danych, ich sprostowanie, usunięcie, ograniczenie przetwarzania, przeniesienie i sprzeciw. Zgodę wycofasz w każdej chwili, pisząc na privacy@cyberpolak.pl. Możesz też złożyć skargę do Prezesa Urzędu Ochrony Danych Osobowych.
Czy musisz podać dane
Nie, ale bez imienia i nazwiska, e-maila i CV nie rozpatrzymy aplikacji. Nie podejmujemy decyzji automatycznie i nie profilujemy kandydatów. CV nie wysyłamy e-mailem.