Biplab Mondal

Key Skills

Software

CVAT

Data Annotation Tech

Google Cloud Vertex AI

Labelbox

Label Studio

Roboflow

Top Subject Matter

No subject matter listed

Top Data Types

Document

Image

Video

Top Task Types

Bounding Box

Computer Programming Coding

Evaluation Rating

Point Key Point

Polygon

Freelancer Overview

As a dedicated AI Training Data Specialist, I bring strong expertise in meticulous data labeling and annotation across various modalities, including images, video, and text. My focus is on delivering high-quality, precise datasets essential for training robust machine learning models. I am proficient in employing various annotation techniques, such as bounding boxes, semantic segmentation, and keypoint labeling for computer vision tasks, and possess a keen eye for detail crucial for nuanced textual annotation and Large Language Model (LLM) evaluation. I am adept at adhering to strict annotation guidelines, ensuring data consistency and accuracy, which are paramount for model performance. My experience includes contributing to projects that require critical thinking to interpret complex data, manage challenging edge cases, and maintain high standards of quality control. I am committed to supporting the development of cutting-edge AI systems by providing the foundational, high-quality data they need to learn and operate effectively.

Labeling Experience

Data labaling and anotation (Imaje, video,Text, Etc)

CVATVideoPoint Key PointClassification

This project focused on developing high-quality, diverse datasets essential for advanced AI model training and evaluation, with a significant emphasis on prompt design for large language models (LLMs) and data annotation for various machine learning tasks. Our primary goal was to create meticulously labeled data to enhance model accuracy, reduce biases, and improve overall performance across multiple modalities. We meticulously annotated a large volume of data, including over 50,000 text prompts for LLM fine-tuning, classifying intent and sentiment, alongside 20,000 images for object detection (bounding boxes) and semantic segmentation, and 500 hours of audio for speech-to-text transcription and emotion recognition. Rigorous quality control measures were implemented, including multi-stage human review, inter-annotator agreement checks, and automated consistency validations, ensuring a >98% accuracy rate. The meticulously labeled data directly contributed to the successful deployment of

This project focused on developing high-quality, diverse datasets essential for advanced AI model training and evaluation, with a significant emphasis on prompt design for large language models (LLMs) and data annotation for various machine learning tasks. Our primary goal was to create meticulously labeled data to enhance model accuracy, reduce biases, and improve overall performance across multiple modalities. We meticulously annotated a large volume of data, including over 50,000 text prompts for LLM fine-tuning, classifying intent and sentiment, alongside 20,000 images for object detection (bounding boxes) and semantic segmentation, and 500 hours of audio for speech-to-text transcription and emotion recognition. Rigorous quality control measures were implemented, including multi-stage human review, inter-annotator agreement checks, and automated consistency validations, ensuring a >98% accuracy rate. The meticulously labeled data directly contributed to the successful deployment of

2024 - 2025

Education

Work History