Open to opportunities

Kushal Jaiswal

@kushaljaiswal

Message

I’m an AI Trainer and LLM Evaluator focused on high-quality localization QA.

India

Message

What I'm looking for

I’m looking for work where I can evaluate and improve LLM quality through rubric-based SBS and localization QA—using clear severity/category/root-cause feedback, consistency checks, and calibration to help teams build safer, more accurate models.

I’m a Hindi-English bilingual AI Trainer and LLM Evaluator specializing in RLHF workflows, SBS response evaluation, and guideline-based quality decisions. In contract work with Outlier AI (Scale AI), I evaluated and ranked 1,000+ AI-generated responses using structured SBS comparisons across reasoning, factuality, instruction-following, tone, and safety—maintaining a 4.8/5 quality score.

I bring a MQM/LQA-style approach to severity, category, and root-cause notes, plus fluency/adequacy/terminology and cultural fit checks for Hindi-English outputs. I also document recurring failure patterns (hallucination, factual drift, over-hedging, instruction misinterpretation) and contribute to clearer, calibration-ready feedback for model improvement.

Experience

Work history, roles, and key accomplishments

Current

AI Trainer / LLM Evaluator

Current

Outlier AI (Scale AI)

Jan 2024 - Present (2 years 6 months)

Evaluated and ranked 1,000+ AI responses using structured SBS comparisons across reasoning, factuality, instruction-following, tone, and safety. Applied guideline-based rubrics for consistent quality decisions, including severity judgment, category selection, and root-cause notes.

RLHF Workflows SBS Comparison MQM LQA Severity Root Cause Analysis Adequacy Evaluation Terminology Consistency Hallucination Detection Factual Grounding Checks Instruction Adherence Evaluation CAT QA Workflow Concepts Prompt Testing Red Teaming

Product Analyst

Independent

Jan 2023 - Jan 2024 (1 year)

Analyzed retention, engagement, and conversion metrics and built KPI dashboards. Used SQL and Power BI to support ongoing performance tracking.

SQL Power BI Retention Analysis KPI Tracking

STEM Subject Matter Expert

Chegg India

Jan 2023 - Jan 2024 (1 year)

Solved 3,000+ Physics and Mathematics problems with step-by-step explanations under strict accuracy and quality guidelines. Maintained reviewer-ready technical education formatting and clear terminology.

Step By Step Explanations Technical Education Content Quality Reviewer Ready Formatting Logical Reasoning Terminology Consistency

AI Voice Data Trainer

Babel Audio

Jan 2023 - Present (3 years 6 months)

Annotated and validated speech datasets for speech recognition and NLP models, achieving a 4.9/5 quality score. Performed transcription QA, consistency checks, and batch-level quality review to improve data reliability.

Speech Recognition Data QA Transcription QA Transcription Consistency Checks Batch Level Quality Review