Hindi Audio Evaluator for AI Research

Contribute to a high-impact audio AI research project by transcribing, annotating, and evaluating Hindi audio and video to help train and benchmark advanced language models. This is a short-term, structured engagement performed remotely, with occasional in-person collaboration in the San Francisco Bay Area when required.

• Listen to and accurately transcribe Hindi audio and video, following detailed formatting, stylistic, and constraint-driven instructions.
• Produce high-quality written outputs in Hindi, and provide supporting work in English when requested.
• Capture linguistic nuance including tone, intent, register, regional expressions, dialectal variation, and contemporary usage.
• Define and document evaluation standards, creating rubrics and grading guidelines in Hindi and English to set expectations for correct and high-quality responses.
• Identify linguistic edge cases such as colloquialisms, grammatical complexities, and code-switching patterns, and document them for consistent reviewer application.
• Run prompts through language models, assess generated outputs, and grade responses for accuracy, completeness, fluency, and clarity of instruction.
• Provide structured feedback to improve model performance on Hindi audio tasks.
• Participate in quality assurance and review cycles to ensure task design, rubrics, and outputs meet the project quality bar, and collaborate with project leads to resolve ambiguities.

• Strong writing, editing, and critical thinking skills.
• Ability to work independently, manage time effectively, and meet deadlines.
• Native or near-native fluency in Hindi, both spoken and written, and professional fluency in English.
• Strong familiarity with spoken Hindi, regional vocabulary, dialects, code-switching including Hindi-English usage, and contemporary language trends.
• Experience accurately transcribing and analysing Hindi audio across general consumer contexts is required.
• Preferred candidates include college students or recent graduates, and those with backgrounds in linguistics, humanities, social sciences, journalism, translation or localisation, or technical disciplines.
• Prior experience with transcription, annotation, evaluation, or research workflows in Hindi, and an interest in AI or language model research, are advantageous.

• Engagement type, hourly work with a short-term, structured project timeline.
• Remote-first role, with occasional in-person collaboration in the San Francisco Bay Area if needed.
• Time commitment of 10 to 20 hours per week.

• $50 per hour.

• Candidates must be based in the San Francisco Bay Area, to enable occasional in-person collaboration.
• Fluency requirements: native or near-native Hindi and professional English.
• Must be able to commit to the stated weekly hours.

• Complete a brief AI-driven interview of approximately 15 minutes.
• Successful candidates will be onboarded and invited to begin project work.