Teams at Realm Labs

Conduct adversarial red teaming of large language models, multimodal models, and agentic systems. Identify vulnerabilities such as jailbreaks, prompt injection, tool-use abuse, and model extraction; automate attack generation and evaluation; measure guardrail robustness; use interpretability methods to locate failure modes; and produce reproducible attacks, evaluations, and technical reports or papers.

Conduct machine learning research and applied ML/software engineering focused on trust and safety in large language models, including prompt injection, safety alignment, privacy, confidentiality, and interpretability. Responsibilities include reviewing literature, formulating research questions, designing experiments, training and evaluating deep models, and communicating findings through technical writing and research papers.

The role involves deploying and optimizing LLMs, designing the ML serving stack, and ensuring high-performance GPU services for production readiness.