
Senior Deep Learning Scientist, AI Security
Posted 1 day ago

Posted 1 day ago
This is a fully remote position, open to applicants in United States.
• Conduct fundamental and applied security research to create, train, fine-tune, and implement large language models for multimodal agentic reasoning systems and protective measures across text, audio, image, video, and tool usage.
• Enhance post-training and alignment techniques, including instruction tuning, preference optimization, and RLHF/RLVR/MOPD.
• Investigate and implement model and system security strategies such as instruction hierarchy, risk detection, and innovative cyber gyms.
• Oversee the collection, development, and benchmarking of multimodal security datasets.
• Guarantee a high standard of evaluation for model accuracy, safety, and success in task completion.
• Collaborate with engineers, data scientists, research teams, and product managers across NVIDIA to expand LLM security and agentic safety solutions.
• Master’s degree (or equivalent experience) or PhD in Computer Science, AI, or Applied Mathematics.
• Over 8 years of relevant professional experience.
• Exceptional programming abilities in Python.
• Strong foundational knowledge in scalable model development.
• Proficiency in deep learning frameworks such as PyTorch.
• Solid understanding of ML/DL techniques and contemporary foundation model architectures, including Transformers and mixture-of-experts models.
• At least 1 year of focused experience or research in one of the core safety domains: LLM Security, Frontier Risks, Agentic Safety, or Multi-turn Security Evaluation.
• Practical experience in post-training multimodal models for multimodal reasoning.
• Research or experience related to LLM backdoors, data poisoning, latent malicious behavior, or structural model vulnerabilities.
• Familiarity with advanced alignment challenges, model deception, manipulation, or loss-of-control scenarios.
• Experience in multi-turn tool-calling, orchestration, and execution risks.
• Knowledge of robust, scalable automated evaluation methodologies for conversational and iterative multi-turn use cases.
• Highly competitive salaries.
• Comprehensive benefits package.
• Equity.
• Benefits for you and your family.
Flare
Helix
Weekday (YC W21)
Prolific
Get handpicked remote jobs straight to your inbox weekly.