AI Safety & Guardrails Data Annotator (Remote)

AI Data Innovations; Pakistan; Remote; PKR 2,295,000 - 3,442,000; Full time

Job summary

AI Data Innovations is seeking technically literate data annotators for an AI safety project focused on guardrail improvements for an enterprise coding assistant. You will review multi-turn conversations, interpret code and logs, and label potential prompt injections and content-moderation risks.

The role requires reading technical content, applying guidelines, and participating in calibration sessions. Training will be provided, and candidates must have a STEM background and strong English

Qualifications

• Bachelor’s degree in Computer Science, Software Engineering or related STEM field or equivalent technical training.
• Ability to read and interpret basic code, configuration files, stack traces and structured outputs.
• Basic understanding of application-security concepts like injection attacks and authentication bypass.
• Professional-level English reading and comprehension skills.
• Strong analytical and contextual reasoning abilities.
• Attention to detail and ability to apply guidelines consistently.

Responsibilities

• Review multi-turn, code-related conversations involving user requests and AI-generated responses.
• Interpret code snippets, system logs, command-line instructions and structured outputs.
• Identify and label potential prompt-injection attempts and hidden malicious content.
• Identify content-moderation risks including hateful or illicit content.
• Evaluate context and intent of technical language beyond individual keywords.
• Replace masked placeholders with neutral synthetic information when required.
• Follow annotation guidelines and participate in calibration sessions.
• Escalate unclear cases according to project procedures.

Skills

Code comprehension

Analytical thinking

English reading

Education

Bachelor’s degree in CS/related STEM

Tools

Label Studio

Prodigy

Job description

AI Data Innovations is seeking technically literate data annotators for an AI safety project focused on guardrail improvements for an enterprise coding assistant. You will review multi-turn conversations, interpret code and logs, and label potential prompt injections and content-moderation risks.

The role requires reading technical content, applying guidelines, and participating in calibration sessions. Training will be provided, and candidates must have a STEM background and strong English