Conversational AI Red Teaming Expert

About OpenTrain

OpenTrain is the #1 platform for finding and building careers in AI training and data labeling. Create a free profile, discover projects that match your skills, and apply in minutes while building a portfolio of work that demonstrates your experience.

• Remote AI training and data-labeling opportunities
• Free profile creation and application process
• A way to build a lasting portfolio in a fast-growing field

About AI Safety and Red Teaming

AI training is the human side of building artificial intelligence. Red teamers help evaluate how models behave under pressure by designing challenging interactions, identifying failures, and documenting risks so systems can be improved.

• Work directly with cutting-edge conversational AI systems
• Use human judgment to find issues automated testing may miss
• Contribute structured findings that expand evaluation coverage

The Role

OpenTrain is seeking a Conversational AI Red Teaming Expert for remote contract work. You will conduct adversarial human evaluation of conversational AI models and agents, probing them for weaknesses and turning your findings into structured data and reproducible artifacts.

This is text-based work that may involve reviewing outputs related to bias, misinformation, or harmful behaviors. Higher-sensitivity assignments are optional, and topics will be communicated before exposure. The role is listed as entry level, while prior experience in adversarial AI work, cybersecurity red teaming, or socio-technical probing is required.

• Remote contractor position
• Part-time engagement with a 20+ hour weekly requirement
• Default commitment of 40 hours per week
• Pay range of $17-$25 per hour
• Worldwide opportunity

What You'll Do

You will test conversational systems through structured adversarial scenarios and communicate both technical and socio-technical risks clearly. Your work will help identify systemic issues and create practical evaluation resources for future testing.

• Probe models and agents for jailbreaks and prompt injections
• Test misuse cases, bias exploitation, and multi-turn manipulation
• Identify vulnerabilities that automated tests may miss
• Annotate failures and classify vulnerabilities
• Flag systemic risks and apply taxonomies, benchmarks, and playbooks
• Document attack cases, datasets, and reports
• Create reproducible artifacts and actionable risk findings
• Expand evaluation coverage for conversational AI

Required Qualifications

This work requires strong adversarial thinking, disciplined testing judgment, and the ability to explain findings to both technical and non-technical audiences. Native fluency in English and Vietnamese is required.

• Prior experience in AI adversarial work, conversational AI red teaming, cybersecurity, or socio-technical probing
• Ability to push AI systems toward failure points
• Ability to identify jailbreaks, prompt injections, bias exploitation, misuse cases, and multi-turn manipulation
• Structured use of taxonomies, benchmarks, or playbooks
• Clear communication of technical and socio-technical risks
• Native fluency in English and Vietnamese

Helpful Background

The following experience may be helpful for this assignment, though it is not presented as a requirement.

• Adversarial machine learning
• Jailbreak datasets
• Prompt injection
• Penetration testing
• Exploit development
• Reverse engineering
• Abuse analysis
• Conversational AI testing
• Psychology, acting, or writing

Why This Work Matters

Every major AI system depends on people who prepare, review, and evaluate data and model behavior. By finding weaknesses in conversational AI and documenting them carefully, red teamers help shape how advanced systems respond in real-world situations.

AI training and data-labeling work can be remote and flexible, making it possible to contribute from anywhere with an internet connection while developing experience in a rapidly growing technology field.

• Work remotely from anywhere
• Choose flexible AI training work that fits your schedule
• Build experience at the intersection of AI safety and human evaluation

How to Apply

Create a free OpenTrain account and build your profile around your AI safety, cybersecurity, language, and evaluation experience. Review the project details and apply in minutes through OpenTrain.

• Apply as an English and Vietnamese speaker
• Highlight relevant red teaming or adversarial testing experience
• Indicate your availability for 20 or more hours per week
• Review sensitivity information before accepting higher-sensitivity assignments