Freelancer

Key Skills

Software

CVAT

Data Annotation Tech

Other

Top Subject Matter

No subject matter listed

Top Data Types

Video

Text

Audio

Top Task Types

Bounding Box

Polyline

Entity (NER) Classification

Classification

Emotion Recognition

Tracking

Freelancer Overview

I am passionate about building high-quality AI solutions, with hands-on experience in data preprocessing, annotation, and validation for machine learning projects. During my internship at IBM, I supported the development of machine learning pipelines, focusing on data quality assessment, feature engineering, and model evaluation. My projects include designing pose estimation systems using MediaPipe for real-time fitness analysis and developing behavioral analytics models for malicious bot detection, both of which required meticulous data labeling and validation to ensure robust model performance. I am proficient in Python, Pandas, NumPy, and Scikit-learn, and have worked extensively with tools like Jupyter and cloud platforms such as Azure and Firebase. My background in computer vision and experience with both supervised and unsupervised learning enable me to contribute effectively to AI training data initiatives, ensuring accuracy and reliability across diverse datasets.

Labeling Experience

Text Classification and Dataset Evaluation for Bot Detection

Data Annotation TechTextEntity Ner Classification

Performed text data labeling and classification for a malicious bot detection system. Labeled behavioral text logs into malicious vs. non-malicious categories. Performed entity extraction and feature tagging. Cleaned and validated raw datasets before model training. Ensured labeling consistency across training and validation datasets. Conducted manual review cycles to improve classification reliability. Worked on structured datasets containing 5,000+ entries.

Performed text data labeling and classification for a malicious bot detection system. Labeled behavioral text logs into malicious vs. non-malicious categories. Performed entity extraction and feature tagging. Cleaned and validated raw datasets before model training. Ensured labeling consistency across training and validation datasets. Conducted manual review cycles to improve classification reliability. Worked on structured datasets containing 5,000+ entries.

2025

Speech Transcription and Emotion Annotation for Conversational AI Training

OtherAudioClassificationEmotion Recognition

Worked on audio dataset annotation for training conversational AI models. Transcribed short audio clips (10–60 seconds) with high accuracy. Cleaned background noise segments and identified unclear speech regions. Annotated speaker emotions (neutral, happy, angry, sad, frustrated). Labeled speaker intent categories (query, complaint, instruction, greeting). Reviewed and corrected ASR (Automatic Speech Recognition) outputs. Conducted quality assurance review cycles to maintain >95% transcription accuracy. Worked on 2,000+ short audio samples. Maintained consistency in punctuation, timestamping, and speaker identification guidelines.

Worked on audio dataset annotation for training conversational AI models. Transcribed short audio clips (10–60 seconds) with high accuracy. Cleaned background noise segments and identified unclear speech regions. Annotated speaker emotions (neutral, happy, angry, sad, frustrated). Labeled speaker intent categories (query, complaint, instruction, greeting). Reviewed and corrected ASR (Automatic Speech Recognition) outputs. Conducted quality assurance review cycles to maintain >95% transcription accuracy. Worked on 2,000+ short audio samples. Maintained consistency in punctuation, timestamping, and speaker identification guidelines.

2025 - 2025

Human Pose Keypoint Annotation for AI Fitness Monitoring

CVATVideoBounding BoxPolyline

Annotated and validated human pose keypoints for an AI-based fitness monitoring system. Labeled body joints (shoulder, elbow, knee, hip, etc.) using keypoint annotation techniques. Reviewed and corrected pose detection outputs for accuracy and consistency. Categorized exercise movements for action recognition (squats, push-ups, etc.). Validated angle measurements between joints for repetition detection. Conducted quality checks to ensure labeling consistency across datasets. Worked with structured image and short video datasets (~1,000+ samples). Maintained strict annotation guidelines and accuracy standards (~95% review compliance).

Annotated and validated human pose keypoints for an AI-based fitness monitoring system. Labeled body joints (shoulder, elbow, knee, hip, etc.) using keypoint annotation techniques. Reviewed and corrected pose detection outputs for accuracy and consistency. Categorized exercise movements for action recognition (squats, push-ups, etc.). Validated angle measurements between joints for repetition detection. Conducted quality checks to ensure labeling consistency across datasets. Worked with structured image and short video datasets (~1,000+ samples). Maintained strict annotation guidelines and accuracy standards (~95% review compliance).

2024 - 2025

Education

Work History