Varunsandesh Manda

Key Skills

Software

Label Studio

Top Subject Matter

AI/NLP and Computer Vision

AI Model Training & RLHF

Text Annotation & NLP

Top Data Types

Text

Image

Top Task Types

Classification

Freelancer Overview

Over the past year I've worked on annotation and AI training data projects covering text classification, NER, image labeling, and AI response evaluation applying RLHF-style rating criteria to assess outputs for accuracy, helpfulness, and alignment with instructions. I bring a technical edge that most annotators don't: with hands-on experience in Python, SQL, and Pandas, I understand what clean and consistent labeled data actually means for a model's performance. What sets me apart is that I've built data pipelines from scratch not just labeled at the surface level. I'm detail-oriented, deadline-driven, and treat every task like the quality of the final model depends on it, because it does.

Labeling Experience

AI Training Data Annotation & Quality Project

Label StudioTextClassification

I contributed to building training datasets for NLP and computer vision models through text and image annotation projects. My work included sentiment analysis, named entity recognition, intent classification, and rating AI-generated responses for alignment. I maintained high quality and consistency throughout all batches, leveraging established industry tools for best practices. • Annotated 500+ text samples for multiple NLP labeling tasks • Rated AI responses for accuracy, coherence, and instruction following using RLHF guidelines • Labeled image datasets with object bounding boxes and tags for computer vision • Maintained data exports and cross-checked label accuracy to exceed 95% consistency

I contributed to building training datasets for NLP and computer vision models through text and image annotation projects. My work included sentiment analysis, named entity recognition, intent classification, and rating AI-generated responses for alignment. I maintained high quality and consistency throughout all batches, leveraging established industry tools for best practices. • Annotated 500+ text samples for multiple NLP labeling tasks • Rated AI responses for accuracy, coherence, and instruction following using RLHF guidelines • Labeled image datasets with object bounding boxes and tags for computer vision • Maintained data exports and cross-checked label accuracy to exceed 95% consistency

2026 - Present

Education

Work History