Odia Language Specialist – AI Training & Evaluation

Job Type :- Contract
Work Mode :- Remote
Compensation :- Up to ₹1,500/hour

About the Role

We are looking for a skilled Odia Language Specialist (SME) to contribute to AI Training and Evaluation projects. The role involves reviewing AI-generated Odia content, creating high-quality reference content, evaluating language quality and providing detailed feedback to improve AI model performance.

Key Responsibilities
Review and evaluate AI-generated Odia content for accuracy, fluency, grammar, spelling, tone and cultural appropriateness.
Create high-quality Odia prompts, reference answers, rewrites and model responses.
Identify mistranslations, ambiguity, logical inconsistencies and meaning shifts.
Evaluate formal and informal language registers and audience-appropriate communication.
Rate and compare multiple AI responses based on correctness and overall quality.
Perform fact-checking and identify inaccurate or hallucinated information.
Detect unsafe, misleading, biased or culturally inappropriate content.
Provide clear and precise feedback in English.
Help improve AI-generated content to make it more natural, reliable, accurate and culturally relevant.

Requirements
Bachelor’s degree or higher in Odia, English Literature, Linguistics, Communications, Journalism, Translation or a related field.
Native or near-native Odia proficiency, with strong reading and writing skills.
C1 or higher English proficiency.
3+ years of professional experience in writing, editing, translation, localization, journalism, education, content QA or related fields.
Strong knowledge of Odia grammar, spelling, punctuation and writing conventions.
Ability to work with different language registers, regional variations and audiences.
Excellent attention to detail and strong written communication skills.
Ability to work independently in a remote, hourly contractor environment.
Prior experience in AI data training, AI evaluation, annotation, translation QA, editorial QA or rubric-based evaluation is preferred.