Marathi AI Response Evaluator

About OpenTrain

OpenTrain is the #1 platform for people building careers in AI training and data labeling. We help contributors discover projects, build a unified portfolio, and grow durable freelance careers teaching AI how to behave — all from one place.

Work with OpenTrain to join a distributed community shaping how modern AI systems learn from human examples. Projects vary by skill and domain, and many let you choose flexible hours and work fully remote.

About AI training work

AI training (data labeling, RLHF, annotation) is the human side of building intelligent systems. Contributors review, rate, and edit model outputs so models learn correct reasoning, tone, and factual accuracy.

This role focuses on response evaluation: your judgments and written notes will directly inform model improvements and guideline updates used by engineers and annotators.

The role

As a Marathi AI Response Evaluator you will review Marathi-language model outputs and produce clear English-language evaluation notes that identify strengths, weaknesses, and factual issues. Your feedback helps improve reasoning, accuracy, tone, and guideline alignment across model responses.

This is a contract, part-time position with a weekly commitment of 20+ hours. Work is remote and open worldwide.

• Employment type: Contractor, Part-time
• Time requirement: 20+ hours per week
• Work location: Remote, worldwide

What you'll do

You will closely compare model responses, rate quality, and write structured evaluation notes in English that explain why outputs are strong or weak and where factual errors or reasoning gaps occur. You will also assess clarity, tone, completeness, and alignment with provided guidelines.

• Review Marathi AI-generated responses for factual accuracy and quality
• Compare strengths, weaknesses, and areas for improvement across outputs
• Check reasoning, clarity, tone, completeness, and guideline alignment
• Write concise, actionable English evaluation notes that engineers and annotators can use

Requirements

You must meet the core requirements below. We will verify language fluency and writing ability during onboarding.

• Native fluency in Marathi (required)
• Strong English writing and analytical communication skills (required)
• Experience using large language models (required)
• Bachelor’s degree (required)
• Structured analytical thinking; background in research, policy, analytics, linguistics, or engineering is preferred

Preferred experience

Candidates with prior RLHF, model evaluation, or annotation experience will be competitive. Experience producing high-quality written content and making fine-grained qualitative judgments across multiple outputs is especially helpful.

• Prior RLHF, model evaluation, or data annotation experience
• Experience writing or editing high-quality written content
• Comfort making fine-grained qualitative judgments and explaining them clearly

Compensation and logistics

Pay is hourly in USD. The posted pay range is $15–$20 USD per hour (advertised hourly rate: $20). As a contractor you will be paid per the contract terms you accept when hired.

You will work remotely and should be available to perform 20+ hours per week. Specific scheduling is flexible but you must meet quality and throughput expectations.

• Hourly pay: $15–$20 USD (hourlyRate: $20)
• Contractor role — paid according to contract terms
• Remote and flexible schedule; 20+ hours/week

Who should apply

Apply if you are a Marathi-native communicator who enjoys detailed qualitative evaluation and can explain judgments clearly in English. This role suits people who want flexible, cutting-edge work that directly influences how language models behave.

You do not need prior annotation experience if you meet the language, writing, and LLM experience requirements — but prior RLHF or evaluation work is a strong plus.

How it works

If selected you will complete onboarding and validation tasks to confirm language fluency and evaluation ability. After onboarding you will receive tasks through OpenTrain with clear guidelines and examples. Your English evaluation notes should be structured, specific, and actionable.

OpenTrain helps you build a portfolio of completed projects and feedback so you can grow in AI training roles over time.

• Onboarding includes validation tasks
• Tasks come with guidelines and examples
• Deliver concise English evaluation notes explaining each judgment