Hong Tran

Key Skills

Software

OneForma

Other

Top Subject Matter

No subject matter listed

Top Data Types

Audio

Text

Top Label Types

Evaluation Rating

Classification

Freelancer Overview

I am a detail-oriented AI evaluator with hands-on experience in data annotation, natural language labeling, and audio quality assessment for conversational AI and speech recognition models. My work includes evaluating TTS audio for clarity and naturalness, classifying cybersecurity content to support fraud detection, and collecting and validating voice and image data for computer vision and NLP applications. With a strong foundation in psychology and peer support, I excel at interpreting guidelines, providing structured feedback, and ensuring high-quality training data. I am also developing my skills in data analytics, including data cleaning, visualization, and SQL, to further enhance my ability to contribute to AI and machine learning projects.

Labeling Experience

Message Categorization

OtherTextClassification

This project focused on classifying SMS messages to support AI‑driven fraud‑detection and safety systems. Annotators reviewed text messages and categorized them into defined intent groups such as Scam, Spam, Suspicious, Clean, and Unknown. The work involved identifying digital‑fraud indicators including homoglyphs, brand spoofing, misleading phrasing, and malicious link patterns.

This project focused on classifying SMS messages to support AI‑driven fraud‑detection and safety systems. Annotators reviewed text messages and categorized them into defined intent groups such as Scam, Spam, Suspicious, Clean, and Unknown. The work involved identifying digital‑fraud indicators including homoglyphs, brand spoofing, misleading phrasing, and malicious link patterns.

2025

TTS Grader

OneformaAudioEvaluation Rating

This project involved evaluating AI‑generated speech across multiple criteria that changed from batch to batch. Tasks included rating pronunciation, accent accuracy, prosody, pacing, emotional tone, and conversational naturalness. Some batches focused on American‑adjacent accents, while others required assessing multi‑speaker dialogue for realism and human‑like interaction.

This project involved evaluating AI‑generated speech across multiple criteria that changed from batch to batch. Tasks included rating pronunciation, accent accuracy, prosody, pacing, emotional tone, and conversational naturalness. Some batches focused on American‑adjacent accents, while others required assessing multi‑speaker dialogue for realism and human‑like interaction.

2025

Education

Work History