Dipesh Chavan

Key Skills

Software

CVAT

Labelbox

Scale AI

Other

Top Subject Matter

Jewellery products

Data Tagging

Image labelling

Top Data Types

Image

Text

Top Label Types

Bounding Box

Polygon

Freelancer Overview

I bring a well-rounded background in data labeling, AI training, and language services, backed by hands-on experience with organizations such as Outlier AI, FutureBeeAI, and Invisible Technologies. My recent roles have involved annotating and optimizing product data for e-commerce, translating and transcribing regional language content, and evaluating AI-generated outputs for factual accuracy and alignment with safety guidelines. This has equipped me with a strong understanding of quality assurance, linguistic nuance, and data annotation best practices. What sets me apart is my ability to work across diverse domains—from customer service and data entry to NLP tasks like sentence translation and audio transcription in Marathi. My technical foundation as an Electronics & Telecom Engineer and early-career experience in RF engineering further support my attention to detail and analytical thinking, essential for high-quality data curation and AI training tasks.

Labeling Experience

Ad Classification & Annotation – Household Product Advertisements

Scale AIVideoClassification

Worked on an ad classification task via the (Uber) Scale AI platform, where the objective was to systematically watch and evaluate video advertisements of various household products such as sunscreen, cleaning products, and skincare items. Each video was assessed based on approximately 25 well-defined parameters, including emotional tone, product visibility, demographic targeting, call-to-action clarity, and visual/audio quality. The goal was to ensure consistent, accurate, and high-quality annotations to train machine learning models for better ad performance analysis and content understanding.

Worked on an ad classification task via the (Uber) Scale AI platform, where the objective was to systematically watch and evaluate video advertisements of various household products such as sunscreen, cleaning products, and skincare items. Each video was assessed based on approximately 25 well-defined parameters, including emotional tone, product visibility, demographic targeting, call-to-action clarity, and visual/audio quality. The goal was to ensure consistent, accurate, and high-quality annotations to train machine learning models for better ad performance analysis and content understanding.

2025 - 2025

AI Assistant Response Quality Comparison

OtherTextEvaluation Rating

Performed side-by-side comparisons of responses generated by two different AI assistants. Evaluated responses based on multiple quality dimensions such as factual accuracy, relevance, clarity, tone, and completeness. Provided final judgments and qualitative feedback to guide improvements in AI model behavior. The task required consistency, critical thinking, and adherence to evaluation guidelines.

Performed side-by-side comparisons of responses generated by two different AI assistants. Evaluated responses based on multiple quality dimensions such as factual accuracy, relevance, clarity, tone, and completeness. Provided final judgments and qualitative feedback to guide improvements in AI model behavior. The task required consistency, critical thinking, and adherence to evaluation guidelines.

2024 - 2024

Education

Work History