Sarath Kumar S

Key Skills

Software

AWS SageMaker

Appen

Telus

Top Subject Matter

No subject matter listed

Top Data Types

Audio

Computer Code Programming

Document

Image

Text

Video

Top Label Types

Bounding Box

Computer Programming Coding

Entity Ner Classification

Object Detection

Point Key Point

Polyline

Prompt Response Writing SFT

Question Answering

Segmentation

Text Generation

Text Summarization

Freelancer Overview

I am a Data Associate with over 3 years of experience delivering high-quality AI training and evaluation data for large-scale systems at Amazon, including Alexa and AWS Bedrock. My expertise spans multimodal data labeling across video, audio, and text, with a strong focus on video annotation involving human movements, gestures, facial expressions, object interactions, and motion events. I have supported computer vision, speech recognition, and smart-home use cases by following detailed annotation guidelines and consistently meeting high quality benchmarks (95%+ accuracy, 98%+ audit quality). In addition to annotation, I have extensive experience in quality assurance and review workflows, including prompt–response evaluation for LLMs, bias and factuality checks, SOP authoring, and reviewer training. I am comfortable working with data annotation tools, refining pre-labeled data, managing ingestion pipelines via Amazon S3, and collaborating closely with QA, linguistics, and engineering teams to deliver production-ready datasets. With strong attention to detail and professional English proficiency, I bring a quality-driven approach to AI training data that emphasizes accuracy, completeness, and consistency.

Labeling Experience

Data Associate

Aws SagemakerVideoBounding BoxPolyline

Worked on large-scale AI training and evaluation projects at Amazon supporting Alexa, Ring, and AWS Bedrock systems. The work involved annotating and reviewing multimodal datasets (video, audio, and text) used for computer vision, speech recognition, and generative AI models. Video annotation tasks included labeling human movements, gestures, facial expressions, object interactions, and motion events, following strict guidelines and quality thresholds. The projects required refining pre-annotated data, ensuring temporal consistency, and maintaining high annotation accuracy through detailed QA and audit workflows. In parallel, contributed to LLM prompt–response evaluation projects focused on correctness, completeness, tone, bias, and policy compliance. Supported end-to-end data operations by managing dataset ingestion via Amazon S3, authoring SOPs, training reviewers, and collaborating with QA and engineering teams to improve tooling and workflows. Across projects, consistently deliver

Worked on large-scale AI training and evaluation projects at Amazon supporting Alexa, Ring, and AWS Bedrock systems. The work involved annotating and reviewing multimodal datasets (video, audio, and text) used for computer vision, speech recognition, and generative AI models. Video annotation tasks included labeling human movements, gestures, facial expressions, object interactions, and motion events, following strict guidelines and quality thresholds. The projects required refining pre-annotated data, ensuring temporal consistency, and maintaining high annotation accuracy through detailed QA and audit workflows. In parallel, contributed to LLM prompt–response evaluation projects focused on correctness, completeness, tone, bias, and policy compliance. Supported end-to-end data operations by managing dataset ingestion via Amazon S3, authoring SOPs, training reviewers, and collaborating with QA and engineering teams to improve tooling and workflows. Across projects, consistently deliver

2022

Education

Work History