English and Bengali AI Red Teaming Specialist

About OpenTrain

OpenTrain is the #1 platform for finding and building careers in AI training and data labeling. Create a profile, discover projects across the industry, and apply in minutes to work that helps shape how modern AI systems behave.

• Free account creation
• Build a portfolio of AI training experience
• Find flexible, remote opportunities in one place

About AI Red Teaming

AI red teaming is part of the human work behind safer, more reliable artificial intelligence. Contributors challenge conversational models with difficult or adversarial scenarios, evaluate their responses, and document weaknesses that automated testing may miss.

• Work directly with cutting-edge conversational AI
• Help identify jailbreaks, prompt injections, bias, and harmful behavior
• Use human judgment to improve model safety and quality

The Role

OpenTrain is hiring an English and Bengali AI Red Teaming Specialist for text-based evaluation of conversational AI models and agents. You will generate human evaluation data by testing adversarial inputs, reviewing model outputs, identifying vulnerabilities, and creating structured artifacts that strengthen AI safety.

This is an entry-level contractor role requiring 20+ hours per week. The pay range is $16-$22 USD per hour, and the role is available to applicants in the listed countries.

• Employment type: Contractor, part-time
• Experience level: Entry level
• Time requirement: 20+ hours per week
• Pay: $16-$22 USD per hour
• Work format: Text-based
• Languages: Native English and Bengali fluency

What You'll Do

You will test conversational models and agents systematically, assess the quality and safety of their responses, and communicate findings clearly to support model improvement.

• Probe models with jailbreaks, prompt injections, misuse scenarios, bias exploitation, and multi-turn manipulation.
• Review text outputs for accuracy, completeness, appropriateness, misinformation, bias, harmful behavior, and systemic risks.
• Annotate failures and classify vulnerabilities using taxonomies, benchmarks, and testing playbooks.
• Create reproducible reports, datasets, and attack cases that communicate findings clearly.
• Identify new scenarios and subtle issues that automated testing may overlook.
• Evaluate AI responses for accuracy, completeness, appropriateness, bias, and harmful behavior.
• Communicate technical and non-technical risk findings clearly.

Required Skills and Evaluation

You should bring strong language judgment, careful attention to subtle issues, and the ability to follow structured quality standards consistently. The selection process will assess your English and Bengali fluency, your understanding of conversational AI risks, and your ability to document findings.

• Native fluency in English and Bengali
• Strong judgment when evaluating language, content, accuracy, completeness, and appropriateness
• Careful attention to errors, inconsistencies, gaps, bias, and potential harms
• Ability to follow structured guidelines and quality standards
• Clear written communication for technical and non-technical audiences
• Adaptability across projects, task types, and customer use cases
• Understanding of jailbreaks, prompt injection, misuse cases, and multi-turn manipulation
• Ability to classify vulnerabilities with taxonomies and document reproducible attack cases

Helpful Background

Experience in any of the following areas can be valuable: adversarial machine learning, jailbreak datasets, prompt injection, RLHF or DPO attacks, model extraction, penetration testing, exploit development, reverse engineering, harassment or disinformation analysis, abuse analysis, conversational AI testing, psychology, acting, or unconventional creative writing.

Higher-sensitivity projects include clear topic guidance and wellness resources.

• Adversarial machine learning or conversational AI testing
• Jailbreak, prompt injection, or model safety research
• Penetration testing, exploit development, or reverse engineering
• Harassment, disinformation, or abuse analysis
• Psychology, acting, or unconventional creative writing

Who Can Apply

This opportunity is open to applicants located in the following countries: AT, BE, BG, CA, CY, CZ, DE, DK, EE, ES, FI, FR, GB, GR, HR, HU, IE, IT, LT, LU, LV, MT, NL, PL, PT, RO, SE, SI, SK, and US.

• Native English and Bengali fluency is required
• Entry-level applicants are welcome
• A commitment of 20+ hours per week is required
• The role is suitable for candidates who can assess sensitive content carefully

Build Your AI Training Career

AI training and data labeling work gives people a flexible way to contribute to the development of modern AI. OpenTrain helps you build a credible profile, show your experience, discover projects that match your skills, and grow toward a long-term career in this fast-moving field.

Create a free OpenTrain account to build your profile and apply in minutes.

• Remote work with flexible project-based opportunities
• Develop experience in AI safety and model evaluation
• Build a portfolio demonstrating your red teaming and evaluation skills