ML Research Scientist Intern – Dynamo Guard / Dynamo Eval / AgentWarden

Dynamo AI; Greater London; On-site; GBP 21,000 - 32,000; Full time

Job summary

A leading AI research firm in Greater London is seeking an ML Research Scientist Intern to advance AI evaluation methodologies and ensure adversarial robustness. You will conduct exciting research on model safety and contribute to novel techniques in AI risk remediation. Ideal candidates are pursuing a Master’s or PhD in relevant fields and have a strong foundation in machine learning and AI, with experience in generative systems highly valued. Join us to make impactful contributions to secure AI practices.

Qualifications

• Currently pursuing a Master’s or PhD in Machine Learning, Artificial Intelligence, or Computer Science.
• Strong theoretical foundation in ML and NLP.
• Experience with large language models or generative AI systems.
• Familiarity with adversarial ML, AI safety, or model evaluation frameworks is a strong plus.
• Demonstrated ability to design rigorous experiments and analyze results critically.

Responsibilities

• Conduct research on LLM evaluation methodologies and adversarial attack generation.
• Develop techniques for detecting hallucinations and policy violations.
• Design experiments to evaluate AI systems under real-world constraints.
• Contribute to research artifacts, including internal technical reports and benchmarking frameworks.
• Collaborate with engineering teams to transition research into deployable guardrails and runtime protections.
• Analyze large-scale model behavior data to uncover vulnerabilities and improvement opportunities.

Skills

Machine Learning

Natural Language Processing

Deep Learning

Adversarial ML

AI Safety

Education

Master's or PhD in Machine Learning, AI, Computer Science, or related field

Tools

Python

PyTorch

Job description

At Dynamo AI, an ML Research Scientist Intern will focus on advancing the state of AI evaluation, adversarial robustness, and agent security. You will contribute to novel research in model safety, hallucination detection, red-teaming, runtime guardrails, and AI risk remediation across DynamoGuard, DynamoEval, and AgentWarden.

Responsibilities

• Conduct research on LLM evaluation methodologies, adversarial attack generation, and model robustness.
• Develop novel techniques for detecting hallucinations, policy violations, prompt injection, and agent misalignment.
• Design experiments to evaluate AI systems under real-world enterprise constraints.
• Contribute to research artifacts, including internal technical reports, benchmarking frameworks, and potentially publications.
• Collaborate with engineering teams to transition research innovations into deployable guardrails and runtime protections.
• Analyze large-scale model behavior data to uncover systematic vulnerabilities and improvement opportunities.

Qualifications

• Currently pursuing a Master’s or PhD in Machine Learning, Artificial Intelligence, Computer Science, or related field.
• Strong theoretical foundation in ML, NLP, or deep learning.
• Experience working with large language models or generative AI systems.
• Familiarity with adversarial ML, AI safety, or model evaluation frameworks is a strong plus.
• Demonstrated ability to design rigorous experiments and analyze results critically.
• Passion for advancing secure and production-grade AI systems.