Open to opportunities

Isabella Castello

@isabellacastello

Message

I evaluate LLM outputs for Brazilian Portuguese safety, quality, and cultural relevance.

Portugal

Message

What I'm looking for

I'm looking to continue evaluating and improving AI systems, especially in RLHF, Trust and Safety, content quality, and PT-BR localization, where I can apply human-centered analysis, policy judgment, and linguistic accuracy.

At Accenture Portugal, I train and evaluate Large Language Models for a major social media platform through RLHF, improving output accuracy, safety, and cultural relevance for Brazilian Portuguese users. I review AI responses for grammar, style, policy compliance, bias, hallucinations, and natural language quality.

I've also investigated complex content issues as a Trust and Safety Analyst and contributed to policy improvements through structured reporting. My Psychology studies, native PT-BR, C2 English, and community support background help me bring a human-centered perspective to AI evaluation.

Experience

Work history, roles, and key accomplishments

Current

LLM Data Analyst

Current

Accenture Portugal

Jan 2026 - Present (8 months)

Performs RLHF to train large language models for a major social media platform, optimizing outputs for accuracy, safety, and cultural relevance in the Brazilian market. Evaluates and refines AI-generated responses for stylistic consistency, grammatical precision, and adherence to client guidelines.

RLHF AI Output Evaluation Quality Analysis Trust and Safety Data Analysis Microsoft Office Google Drive