Remote AI Safety Red Teaming Expert

OpenTrain AI; Österreich; Vor Ort; EUR 19.000 - 26.000; Teilzeit

Zusammenfassung

OpenTrain AI is seeking an AI Red Teaming Expert to probe conversational models and agents with adversarial inputs. You will uncover vulnerabilities and produce human-generated data to support AI safety improvements.

This remote, part-time contractor role focuses on jailbreaks, prompt injections, misuse cases, and bias exploitation. Fluency in English and Malayalam is required. The position offers flexible scheduling and collaboration with safety teams for 20+ hours weekly.

Qualifikationen

• Experience with jailbreaks, prompt injection, or adversarial AI testing.
• Ability to follow taxonomies and quality standards.
• Strong written communication and careful judgment.
• Native fluency in English and Malayalam.

Aufgaben

• Probe conversational AI systems for jailbreaks and prompt injections.
• Review outputs involving bias and safety concerns.
• Annotate failures and classify vulnerabilities using taxonomies.
• Identify systemic risks and gaps that automated tests may miss.
• Create reproducible reports, datasets, and attack cases.
• Expand evaluation coverage across projects and task types.

Kenntnisse

Jailbreaks

Prompt injection

Adversarial testing

Bias analysis

Jobbeschreibung

OpenTrain AI is seeking an AI Red Teaming Expert to probe conversational models and agents with adversarial inputs. You will uncover vulnerabilities and produce human-generated data to support AI safety improvements.

This remote, part-time contractor role focuses on jailbreaks, prompt injections, misuse cases, and bias exploitation. Fluency in English and Malayalam is required. The position offers flexible scheduling and collaboration with safety teams for 20+ hours weekly.