Kairav Mittal

Key Skills

Software

Data Annotation Tech

Label Studio

Top Subject Matter

No subject matter listed

Top Data Types

Audio

Image

Text

Top Task Types

Classification

Evaluation Rating

Question Answering

Translation Localization

Freelancer Overview

I have experience in data labeling and AI training workflows, with a focus on LLM evaluation, prompt rating, localization, and image annotation. My background includes work on both text and image tasks, including image classification, bounding box annotation, and LLM response evaluation for quality, relevance, and tone. I’ve also contributed to multilingual localization efforts, helping align LLM outputs across different languages and cultural contexts. Having lived in Taiwan from childhood, I bring a strong understanding of Mandarin Chinese and cross-cultural understanding. I'm well versed in English, Chinese, and Spanish, which has allowed me to work in diverse datasets. My work is sincere and reflects deep attention to detail, critical thinking about model behavior, and a technical foundation of coding that supports my overall thinking.

Labeling Experience

LLM Localization and Evaluation

Data Annotation TechTextTranslation LocalizationEvaluation Rating

I worked on localizing outputs from a large language model, making sure that responses generated in English and Chinese were accurate, culturally appropriate, and aligned in meaning and tone. The job wasn’t just translating — it involved reviewing how well the model understood context, adapted idioms, and handled region-specific references. Sometimes, I'd rewrite parts of a response or flag examples where the model misunderstood the cultural nuance.

I worked on localizing outputs from a large language model, making sure that responses generated in English and Chinese were accurate, culturally appropriate, and aligned in meaning and tone. The job wasn’t just translating — it involved reviewing how well the model understood context, adapted idioms, and handled region-specific references. Sometimes, I'd rewrite parts of a response or flag examples where the model misunderstood the cultural nuance.

2024 - 2024

Image Classification English or Hindi

Data Annotation TechImageClassification

I worked on a project classifying images based on their content using labels in either English or Hindi, depending on the context. The task involved looking closely at images and deciding the most accurate label from a set of options — sometimes simple, sometimes needing more cultural or contextual understanding. Since the data included both languages, I had to switch between them fluidly and make sure the labels fit the meaning in both cases.

I worked on a project classifying images based on their content using labels in either English or Hindi, depending on the context. The task involved looking closely at images and deciding the most accurate label from a set of options — sometimes simple, sometimes needing more cultural or contextual understanding. Since the data included both languages, I had to switch between them fluidly and make sure the labels fit the meaning in both cases.

2023 - 2023

Education

Work History