Jagdish Arora

Key Skills

Software

Internal/Proprietary Tooling

Appen

Labelbox

Top Subject Matter

LLMs training, Translation/Localization, Image recognition, labelling and classification.

LLM Evaluation in English, Hindi and Marathi

Satellite image classification.

Top Data Types

Text

Image

Audio

Top Label Types

Entity Ner Classification

Text Generation

Text Summarization

Bounding Box

Prompt Response Writing SFT

Function Calling

Freelancer Overview

With over 15 years of diverse experience, I specialize in AI data evaluation, annotation, and training data preparation for leading platforms such as Amazon, Microsoft, Uber, and Centific. My work includes evaluating LLM queries, performing search engine and ad analysis, and conducting extensive patent research. I am skilled in data labeling for NLP tasks, audio and video transcription, and quality analysis, with hands-on experience in medical, e-commerce, and general content domains. I am proficient with CAT tools (MemoQ, Trados, Cafe Tran, MateCat, Memsource), digital advertising platforms, and various QA and testing tools. My background also includes translating and proofreading large volumes of content in Hindi, Marathi, and English, as well as participating in projects involving AI model improvement, app and web application testing, and hardware/software QA. I am committed to delivering high-quality, accurate training data and have a proven track record of successfully completing complex AI and data projects.

Labeling Experience

Labelbox/Alignerr

LabelboxAudioFunction Calling

Generate Audio transcripts and codes in JSON and Python and evaluate the results and Function calling by the models.

Generate Audio transcripts and codes in JSON and Python and evaluate the results and Function calling by the models.

2025

LLM Training comparison of different LLM outputs

Internal Proprietary ToolingTextPrompt Response Writing SFT

I have to evaluate the best prompt and response from a set of three responses given by three different LLMs. The prompt/responses may either be in English or Hindi languages. Also, in some cases, I have to check if the prompt instructions have been followed correctly in the responses given by the three LLM options and also give my reasons and the defects found in them.

I have to evaluate the best prompt and response from a set of three responses given by three different LLMs. The prompt/responses may either be in English or Hindi languages. Also, in some cases, I have to check if the prompt instructions have been followed correctly in the responses given by the three LLM options and also give my reasons and the defects found in them.

2024

Amazon Claude LLM

Internal Proprietary ToolingTextEntity Ner ClassificationText Generation

I was involved in Amazon Claude LLM training project(ongoing). This project involved checking product data generated by Amazon LLM, classification of the products, editing of the product descriptions, doing QA of work done by others and various related tasks. The project is ongoing and work comes after certain intervals. I am also doing similar work for another LLM.

I was involved in Amazon Claude LLM training project(ongoing). This project involved checking product data generated by Amazon LLM, classification of the products, editing of the product descriptions, doing QA of work done by others and various related tasks. The project is ongoing and work comes after certain intervals. I am also doing similar work for another LLM.

2024

Image Anotation

AppenImageBounding BoxEntity Ner Classification

I had to label various movements of people moving around in the city and note the steps taken and the movements done (left, right, center etc.) and do labelling of all the objects and scenes in them.

I had to label various movements of people moving around in the city and note the steps taken and the movements done (left, right, center etc.) and do labelling of all the objects and scenes in them.

2020 - 2020

Education

Work History