Senior ML Engineer, AI Security
Summary
Build and harden ML models that detect prompt-injection and other AI security threats across languages and data types, using NLP and vision techniques in production.
Example projects
Extend the scope of one of our detectors to support multiple languages without dropping performance on already supported languages.
Increase the scope where our defenses perform by 100x (e.g., multi-lingual data, code, etc).
Extend modeling and data generation to enable passing additional context to our prompt injection detectors, such as the LLM output and behavioral metadata.
Red team our model to understand the most pressing vulnerabilities of our detectors and prioritize the data collection and generation required to fix them.
What you’ll bring
You have at least 4 years of experience shipping machine learning models to production, ideally in environments where reliability and quality are key (e.g., autonomous driving, industrial applications, healthcare).
Most of your experience productionizing ML models has been in unstructured data spaces such as NLP and vision.
You have had experience working on end-to-end development of ML models.
You have experience working on the “long tail” of data distribution, bringing models from prototypes to real-world systems.
You aspire to play a pivotal role in defining the future of secure AI within a mission-driven company with ambitious goals.