AI Security Researcher
TypeRemote job
LocationKorntal-Münchingen
Posted1 hour ago
About the Role
As an AI Security Researcher, you bring deep expertise in AI/ML security, software development in Python, backed by a minimum MSc in Machine Learning, Computer Science, or related discipline. You have hands-on experience with AI workloads, large language models, training frameworks, advanced optimization algorithms, GPU clusters, and ideally SLURM in HPC environments. You excel at modeling AI threats—model/data poisoning, permission escalation, backdoors, data drift attacks, prompt injection—and translating them into practical security defenses.
You will be one of Haste's first AI security researchers pioneering defenses for trillion-parameter AI training environments alongside HPC and quantum workloads.
Key Responsibilities
Adversarial Threat Research: Investigating emerging attack vectors and developing techniques to uncover vulnerabilities in AI models, data pipelines, and infrastructure.
Threat Modeling & Risk Analysis: Mapping and characterizing potential threats across the AI lifecycle—from data ingestion to deployment—to prioritize security investments.
Developing Defenses & Guardrails: Designing and implementing detection logic and researching mitigation methods to AI threats.
Collaborating with Engineering Teams: Translating research findings into actionable, production-grade security requirements for product teams, ensuring threat detection makes sense.
Documentation & Reporting: Creating technical reports, whitepapers, and proof-of-concept (PoC) exploits to document vulnerabilities and share findings within the organization or to the public.
Monitoring AI Ecosystem Trends: Staying updated on the latest AI security standards, frameworks (e.g., OWASP Top 10 for LLMs, MITRE ATLAS), and academic research.
Requirements
University degree in AI/ML or relevant discipline.
Current residence in Germany or Austria.
Full-time availability from December 1, 2026, remote from Germany or Austria only.
Find more English Speaking Jobs in Germany on Arbeitnow
As an AI Security Researcher, you bring deep expertise in AI/ML security, software development in Python, backed by a minimum MSc in Machine Learning, Computer Science, or related discipline. You have hands-on experience with AI workloads, large language models, training frameworks, advanced optimization algorithms, GPU clusters, and ideally SLURM in HPC environments. You excel at modeling AI threats—model/data poisoning, permission escalation, backdoors, data drift attacks, prompt injection—and translating them into practical security defenses.
You will be one of Haste's first AI security researchers pioneering defenses for trillion-parameter AI training environments alongside HPC and quantum workloads.
Key Responsibilities
Adversarial Threat Research: Investigating emerging attack vectors and developing techniques to uncover vulnerabilities in AI models, data pipelines, and infrastructure.
Threat Modeling & Risk Analysis: Mapping and characterizing potential threats across the AI lifecycle—from data ingestion to deployment—to prioritize security investments.
Developing Defenses & Guardrails: Designing and implementing detection logic and researching mitigation methods to AI threats.
Collaborating with Engineering Teams: Translating research findings into actionable, production-grade security requirements for product teams, ensuring threat detection makes sense.
Documentation & Reporting: Creating technical reports, whitepapers, and proof-of-concept (PoC) exploits to document vulnerabilities and share findings within the organization or to the public.
Monitoring AI Ecosystem Trends: Staying updated on the latest AI security standards, frameworks (e.g., OWASP Top 10 for LLMs, MITRE ATLAS), and academic research.
Requirements
University degree in AI/ML or relevant discipline.
Current residence in Germany or Austria.
Full-time availability from December 1, 2026, remote from Germany or Austria only.
Find more English Speaking Jobs in Germany on Arbeitnow
Apply on Arbeitnow →
Job sourced from Arbeitnow. Applications happen directly on the original platform — we never collect your data.