Computational Linguist

About the position

As a Computational Linguist at SnowCrash Labs, you will bring a deep understanding of language, meaning, and communication to our AI security platform. Your expertise will be critical for detecting the subtle linguistic patterns that distinguish legitimate AI interactions from adversarial manipulation, deception, and social engineering attempts.

Job responsibilities

• Analyze and categorize the linguistic patterns used in adversarial attacks against AI systems, including prompt injection, goal manipulation, and deceptive communication.
• Develop natural language understanding models that detect manipulation patterns such as flattery-based influence, fear-based pressure, and strategic misdirection in AI interactions.
• Build evaluation frameworks that assess AI model outputs for deception signals, selective disclosure, and other forms of dishonest communication.
• Collaborate with the research team to expand our cognitive risk taxonomy, identifying new categories of linguistic manipulation as AI systems become more sophisticated.
• Create training datasets and evaluation benchmarks for adversarial language detection across multiple languages and domains.

This role is perfect for someone who understands that the most dangerous AI attacks are not technical exploits but sophisticated uses of language to manipulate behavior.