LLM-Safety Research Intern

IN-CYPHER, Imperial College London’s flagship research program in Singapore, advances security, privacy, and trust in healthcare—from implantables and wearables to hospital networks and clinical workflows. This internship will be about evaluating LLM jailbreak transferability and defence robustness in high-stakes healthcare contexts.

Key Responsibilities:

• Tests jailbreak transferability across major LLM APIs
• Quantifies defense robustness under healthcare-specific threat models
• Benchmarks both general-purpose and medical LLMs
• Surfaces insights for IN-CYPHER-aligned deployment guidelines

Minimum Requirements:

• Currently enrolled in an undergraduate program in Computer Science/Engineering.
• Experience with LLMs/NLP and/or AI security (e.g., privacy, robustness)
• Python (requests/asyncio, logging), Deep learning frameworks (e.g., PyTorch) , pandas, Git, Docker
• Experience with web APIs, experiment hygiene
• Reading and familiarity with ML safety papers, adversarial prompting, LLM APIs
• Clear technical writing + repo reproducibility standards

Stipend: $1,500 per month.

Duration: 6–9 months

Informal enquiries are greatly welcome and can be directed to Dr Viktor Schlegel v.schlegel@imperial.ac.uk.

Questions about the recruitment process, should go to the HR at Imperial Global Singapore hrigsingapore@imperial.ac.uk.