The Role
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.
Summary Generated by Built In
Role Overview
Expected Background: Adversarial ML and Red Teaming
Expected Background: ML
Expected Background: Software Engineering
Compensation & Benefits
Requirements
- You will try to break the systems we build and the models we protect, and turn what you find into something the team can act on: a reproducible attack, an evaluation that catches it, and a written account of why it works.
- Expect some mix of: eliciting unsafe behaviour from aligned LLMs and multi-modal models; prompt injection and tool-use abuse against agentic systems; automating attack generation and evaluation rather than hand-crafting one-off prompts; and measuring whether guardrails hold under pressure. Where RealmLabs' interpretability work gives you access to a model's internals, use it.
- We aim for a paper or public technical report out of every internship, plus attacks that stay in our evaluation suite after you leave.
- MUST HAVE prior red-teaming or pen-testing experience like DefCon CTFs or bugbounties. General SWE experience is not suitable for this role.
- MUST HAVE Familiarity with agentic systems and their attack surface — tool calls, retrieval, memory, multi-agent orchestration.
- Hands-on experience attacking or stress-testing models, from any direction: jailbreaks, prompt injection, adversarial examples, data poisoning, model extraction, or evaluating safety and moderation systems.
- Able to read a paper and implement its attack.
Expected Background: ML
- Machine learning tools: pytorch, huggingface, transformers, datasets.
- Applied deep learning and LLM experience.
- Training and evaluating deep models.
- (nice to have) finetuning LLMs, multi-modal LLMs.
- (nice to have) Familiarity with ML[NLP,LLM,Vision] interpretability methods, sparse autoencoders, linear probes — as a way of locating failure modes, not as an end in itself.
- Development environments and tools:
- unix, git, basic clouds usage on AWS and/or GCP
- jupyter
- Programming:
- python
- (nice to have) “programming languages well-roundedness”
- experience in statically-typed and functional languages
- Market aligned compensation for interns in the bay area.
- Must be authorized to work in the USA or must be able to obtain CPT (Curricular Practical Training) approval from host university.
Skills Required
- Prior red-teaming or penetration-testing experience, such as DefCon CTFs or bug bounties
- Familiarity with agentic systems and attack surfaces involving tool calls, retrieval, memory, and multi-agent orchestration
- Hands-on experience attacking or stress-testing machine learning models or safety and moderation systems
- Ability to read research papers and implement published attacks
- Experience with PyTorch, Hugging Face, Transformers, and Datasets
- Applied deep learning and large language model experience, including training and evaluating deep models
- Experience with Unix, Git, basic AWS or GCP usage, and Jupyter
- Python programming experience
- Familiarity with fine-tuning language models or multimodal language models
- Familiarity with machine learning interpretability, sparse autoencoders, or linear probes
- Well-rounded programming language experience, including statically typed and functional languages
- Authorization to work in the USA or ability to obtain CPT approval from a host university
Am I A Good Fit?
Get Personalized Job Insights.
Our AI-powered fit analysis compares your resume with a job listing so you know if your skills & experience align.
Success! Refresh the page to see how your skills align with this role.
The Company
What We Do
Realm Labs is an AI security company focused on helping enterprises secure and monitor AI applications and data. It develops an AI-based authorization platform designed to prevent data leaks and enable secure generative AI usage with features like agent security and guardrails.
