Harsh K.

Key Skills

Software

Don't disclose

Top Subject Matter

Large language model response evaluation

safety/policy review

and content quality assurance.

Top Data Types

Text

Audio

Document

Top Task Types

Prompt + Response Writing (SFT)

Evaluation/Rating

Freelancer Overview

AI Trainer — LLM Response Evaluation (Invisible AI). Brings 4+ years of professional experience across complex professional workflows, research, and quality-focused execution. Core strengths include Other. Education includes Integrated Bachelor of Science and Master of Science, IISER Bhopal (2023). AI-training focus includes data types such as Text and labeling workflows including Evaluation and Rating.

Labeling Experience

AI Trainer — LLM Response Evaluation (Invisible AI)

OtherText

Responsible for rubric-based evaluation of LLM responses across accuracy, relevance, clarity, and safety. Produced concise, well-reasoned justifications for each rating to support consistent model feedback. Conducted ranking and comparison across model outputs to generate human preference data for training and quality improvement. • Rubric-driven scoring and written justification • Ranking/comparison for human preference data • Safety and policy violation flagging with guideline adherence • Text labeling, search quality rating, audio transcription/segmentation, and translation/localization QA

2025 - 2026

AI Trainer — LLM Response Evaluation (Outlier AI)

OtherText

Performed LLM response and prompt evaluation using platform guidelines and detailed rubrics. Completed labeling and classification tasks while handling edge cases, ambiguous inputs, and strict guideline compliance. Conducted safety and policy review for borderline cases with documented, reasoned decisions. • LLM response evaluation and prompt evaluation • Labeling and classification with edge-case handling • Safety/policy review and documented decisions • Content quality review, audio transcription, and localization QA

2024 - 2025

Data Analyst Intern (AI Variant)

OtherText

Worked with structured datasets to support rubric-aligned analysis and evidence-backed written summaries. Identified patterns, inconsistencies, and outliers using systematic review methods consistent with QA and guideline-compliance processes. Transformed analyzed data into clear written outputs that align with evaluation and feedback workflows. • Structured dataset analysis for QA • Evidence-backed written summaries • Outlier and inconsistency detection • Quality-focused review aligned with rubric workflows

2023 - 2024

Education

Work History