Lovely Cris Cagas

Key Skills

Software

Appen

Top Subject Matter

No subject matter listed

Top Data Types

Audio

Computer Code Programming

Document

Image

Top Label Types

Audio Recording

Data Collection

Question Answering

Text Generation

Text Summarization

Translation Localization

Freelancer Overview

I am an AI training and data labeling professional with hands-on experience supporting large language model (LLM) development through scenario evaluation, AI conversation refinement, and structured data annotation. Through Appen projects, I have reviewed and improved multi-turn AI conversations, defined gold-standard behaviors, and applied QA-style analysis to identify inconsistencies, ambiguities, and edge cases. I am comfortable working with structured formats such as JSON, validating schema accuracy, and flagging missing or unclear information to ensure high-quality training datasets. In addition to AI training work, I bring a strong professional writing and editorial background as a Campus Journalist and former Editor-in-Chief, which strengthens my attention to detail, analytical thinking, and quality ownership. I have experience extracting and validating information from complex documents, collaborating with remote QA teams, and refining outputs to meet strict evaluation guidelines. These combined skills allow me to contribute reliable, accurate, and well-documented training data that improves AI model performance and evaluation consistency.

Labeling Experience

Virtual Assistant / AI Evaluation Specialist

AppenImageQuestion AnsweringText Generation

Contributed to LLM agent evaluation by designing realistic, reusable scenarios and defining gold-standard behaviors, acceptable alternatives, and edge cases. Reviewed, annotated, and refined multi-turn agent outputs to support consistent benchmarking, improving coherence, accuracy, and naturalness in English and Tagalog. Built structured ground-truth datasets by extracting financial data from anonymized documents into validated JSON schemas, flagging ambiguities. Applied strong QA and analytical thinking to identify inconsistencies and collaborate with cross-functional teams to improve evaluation frameworks.

Contributed to LLM agent evaluation by designing realistic, reusable scenarios and defining gold-standard behaviors, acceptable alternatives, and edge cases. Reviewed, annotated, and refined multi-turn agent outputs to support consistent benchmarking, improving coherence, accuracy, and naturalness in English and Tagalog. Built structured ground-truth datasets by extracting financial data from anonymized documents into validated JSON schemas, flagging ambiguities. Applied strong QA and analytical thinking to identify inconsistencies and collaborate with cross-functional teams to improve evaluation frameworks.

2025 - 2025

Education

Work History