Manish Kumar

Key Skills

Software

Labelbox

Top Subject Matter

Large Language Models

AI Safety

Multi-lingual Conversational AI

Top Data Types

Text

Image

Top Task Types

RLHF

Classification

Freelancer Overview

I have extensive experience as an AI Trainer and Content Specialist, focusing on the end-to-end development of high-quality training datasets for Large Language Models (LLMs). My work centers on **RLHF (Reinforcement Learning from Human Feedback)**, where I specialize in ranking model outputs, identifying subtle hallucinations, and performing expert-level red-teaming to ensure AI safety and alignment. I have successfully led projects involving complex **Chain-of-Thought (CoT)** prompting and SFT (Supervised Fine-Tuning) data curation, which directly improved model reasoning capabilities in technical domains like mathematics and coding. What sets me apart is my ability to bridge the gap between human intuition and machine learning requirements. I am highly proficient in developing nuanced rating rubrics that capture granular details like tone, factuality, and instruction-following. My technical background in Python and data engineering allows me to automate quality assurance processes, significantly reducing error rates in massive datasets. Whether it is fine-tuning models for creative writing or optimizing them for rigorous factual accuracy, my focus remains on producing diverse, bias-free, and high-impact data that drives the next generation of AI performance.

Labeling Experience

Senior AI Content Specialist & Trainer

TextRLHF

As Senior AI Content Specialist & Trainer, I led a team to optimize AI responses and improve model performance using RLHF and custom evaluation rubrics. I collaborated with machine learning engineers to turn human-centric feedback into actionable datasets for large language models. My role also involved designing red-teaming protocols to assess and mitigate AI risks. • Led optimization of conversational AI in 5+ languages. • Engineered proprietary evaluation metrics for factuality and safety. • Oversaw and contributed to human feedback data collection and labeling. • Conducted adversarial testing to identify and address model biases.

As Senior AI Content Specialist & Trainer, I led a team to optimize AI responses and improve model performance using RLHF and custom evaluation rubrics. I collaborated with machine learning engineers to turn human-centric feedback into actionable datasets for large language models. My role also involved designing red-teaming protocols to assess and mitigate AI risks. • Led optimization of conversational AI in 5+ languages. • Engineered proprietary evaluation metrics for factuality and safety. • Oversaw and contributed to human feedback data collection and labeling. • Conducted adversarial testing to identify and address model biases.

2022 - Present

AI Data Associate

LabelboxImageClassification

As an AI Data Associate, I managed end-to-end data annotation projects for both computer vision and NLP models. My contributions included production and management of gold-standard datasets for benchmarking advanced LLMs. I implemented automation scripts to improve data labeling quality and efficiency. • Managed data annotation workflows for images and text. • Authored gold-standard datasets adopted in LLM benchmarking. • Implemented automated QA for reduced labeling error rates. • Worked with teams using industry tools such as Labelbox and Scale AI.

As an AI Data Associate, I managed end-to-end data annotation projects for both computer vision and NLP models. My contributions included production and management of gold-standard datasets for benchmarking advanced LLMs. I implemented automation scripts to improve data labeling quality and efficiency. • Managed data annotation workflows for images and text. • Authored gold-standard datasets adopted in LLM benchmarking. • Implemented automated QA for reduced labeling error rates. • Worked with teams using industry tools such as Labelbox and Scale AI.

2019 - 2021

Education

Work History