b
bru_paiva

Bruna P

@bru_paiva

AI Model Evaluator and Data Analyst

Brasil
Inglês, Português
Algumas informações são exibidas no idioma inglês.
Sobre mim
Bilingual (Portuguese/English) AI model evaluator and data analyst with 3 years evaluating AI-generated content for accuracy, coherence and guideline adherence, and 2 years building analytics pipelines and dashboards. I hold degrees in Language Studies and Dentistry, a postgraduate degree in Data Science and Big Data, and I am studying Software Engineering. Tools: SQL, Python (Pandas, NumPy, Scikit-learn), Power BI, BigQuery, Looker Studio, Jupyter, Git/GitHub. Detail-oriented, autonomous, always on time.... Saiba mais

Habilidades

b
bru_paiva
Bruna P
offline • 

Conheça meus serviços

Anotação e marcação de dados
I will evaluate ai model responses in portuguese and english
A partir deUS$ 65 /projetoMais detalhes
Limpeza de Dados
I will clean your data with python and sql
A partir deUS$ 35 /projetoMais detalhes

Portfólio

Experiência profissional

RWS_Group

AI Model Evaluator

RWS Group • Freelance

May 2023 - Present • 3 yrs 5 mos

Evaluated AI-generated content against technical and factual accuracy guidelines, cross-referencing responses with source guidelines and documenting inconsistencies for model retraining. Rated audio quality (voice naturalness and speaker similarity) using calibrated rubrics. Delivered 100% on time, working autonomously with distributed international teams under fixed SLAs.

Laboratoria

Data analyst

Laboratoria • Meio período

Mar 2024 - Jul 2024 • 4 mos

Completed an intensive 4-month data analytics bootcamp, delivering end-to-end projects. Built a 10-segment RFM customer model (62% of revenue traced to 3 segments) with an interactive Looker Studio dashboard and cohort retention analysis. Analyzed a full month of US flight delays and cancellations, testing 4 hypotheses with chi-square and relative-risk tests in Python and a Power BI + DAX dashboard. Built a loan-default risk classifier (logistic regression, SMOTE, AUC-ROC) that scored customers into 3 risk tiers.