Marathi AI Data Labeler - Onsite, Human-in-the-Loop

Welo Data; Washington; On-site; USD 41,000 - 52,000; Full time

Benefits offered by this job

Paid vacation

Health insurance

401(k) plan

Dental insurance

Vision insurance

HSA

Employee assistance program

Free meals

Commuter benefits

Job summary

Welo Data is seeking a Data Labeling Associate for Marathi speakers to join our onsite team in the United States. You will evaluate AI model outputs, provide structured feedback, and help improve how these systems behave in real‑world scenarios.

This role sits at the intersection of data quality, model evaluation, and human judgment. The position is full-time (40 hours/week) as a W2 employee with full benefits.

Qualifications

• Native Marathi speaker with university degree.
• English proficiency at B2 or higher.
• 1–2 years of professional writing experience with structured writing.
• Ability to apply complex writing rules and guidelines consistently.
• 2+ years of relevant experience in GenAI data safety/quality.

Responsibilities

• Evaluate AI model outputs and provide structured, high-quality feedback.
• Perform audit-based reviews of data and model behavior, identifying patterns and edge cases.
• Apply guidelines thoughtfully and flag when they don’t reflect real-world scenarios.
• Contribute to improving evaluation frameworks beyond execution.
• Identify trends in model performance and communicate insights clearly.
• Participate in team discussions, calibrations, and stakeholder syncs.
• Partner with leads and cross-functional teams to refine quality standards.
• Document findings in a clear, concise, actionable way.

Skills

Marathi language proficiency

English (B2+)

Strong writing skills

Critical thinking

Attention to detail

Communication skills

Education

Bachelor's degree or higher

Job description

Welo Data is seeking a Data Labeling Associate for Marathi speakers to join our onsite team in the United States. You will evaluate AI model outputs, provide structured feedback, and help improve how these systems behave in real‑world scenarios.

This role sits at the intersection of data quality, model evaluation, and human judgment. The position is full-time (40 hours/week) as a W2 employee with full benefits.