Chidera Ezema

Key Skills

Software

Appen

Top Subject Matter

No subject matter listed

Top Data Types

Audio

Document

Geospatial Tiled Imagery

Image

Video

Top Task Types

Segmentation

Cuboid

Question Answering

Emotion Recognition

Object Detection

Freelancer Overview

I am an experienced AI training specialist with over three years in data annotation and NLP, particularly focused on image and text labeling for computer vision and language models. My expertise spans bounding boxes, semantic segmentation, and key-point labeling, with hands-on proficiency in tools like Labelbox, CVAT, Amazon SageMaker Ground Truth, and Scale AI. I have led end-to-end annotation projects, achieving a consistent 98% accuracy rate, and have contributed to the development of inclusive AI by curating and localizing large-scale Igbo language datasets for translation, transcription, and sentiment analysis. My background in RLHF and supervised fine-tuning, combined with a strong commitment to data quality and cultural context, allows me to deliver high-precision, bias-reduced training data for diverse industries including e-commerce and autonomous systems.

Labeling Experience

A13019-Livingstone-E LLM project

AppenVideoSegmentationCuboid

I managed end-to-end video annotation tasks focused on enhancing multimodal AI models. my responsibilities included human-object detection (counting participants) and audio-visual synchronization to map specific dialogue to individual speakers. Additionally, I acted as a subject matter expert in Natural Language Processing (NLP) refinement, correcting erroneous prompts and generating high-quality Q&A datasets to improve the model's contextual understanding and response accuracy.

I managed end-to-end video annotation tasks focused on enhancing multimodal AI models. my responsibilities included human-object detection (counting participants) and audio-visual synchronization to map specific dialogue to individual speakers. Additionally, I acted as a subject matter expert in Natural Language Processing (NLP) refinement, correcting erroneous prompts and generating high-quality Q&A datasets to improve the model's contextual understanding and response accuracy.

2025 - 2025

Education

Work History