NLP Researcher · ML Engineer · Generative AI

Isabela Araujo

I build and research AI systems that work in the real world. As an ML Engineer at Nubank, I operationalize machine learning at scale across Brazil and Mexico. As a researcher at HAILab (PUCPR), I investigate how large language models can be applied to clinical NLP — text normalization, abbreviation disambiguation, and term mapping in Portuguese.

My focus: NLP, LLMs, and generative AI - especially in healthcare contexts where language is messy, multilingual, and high-stakes.


Oct 2022 – Present
Machine Learning Engineer
Nubank · São Paulo, Brazil
Operationalizes ML and AI systems at scale across Brazil and Mexico: deployment pipelines, model monitoring and parametrization, governance integration, and DevOps workflows. Contributes to Data Science platformization and maintains documentation and best practices for model lifecycle management in a global team.
PythonLLMsGenerative AI TensorFlowPyTorchLangChain DatabricksDockerKubernetes
Nov 2023 – Present
NLP Researcher
HAILab – PUCPR · Paraná, Brazil
Research in clinical NLP: abbreviation identification and disambiguation, text normalization, codification, and term mapping using standardized medical terminologies. Explores prompt engineering and LLMs - GPT, Llama, Gemini, and Sabiá (Portuguese model) - applied to real-world clinical narratives.
Clinical NLPLLMsPrompt Engineering Hugging FacePortuguese NLP
Nov 2022 – Aug 2023
NLP Researcher
Comsentimento · São Paulo, Brazil
Few-shot and zero-shot learning on anonymized medical datasets. Built NLP pipelines for text classification, NER, and prompt-based tasks in clinical and sensitive data contexts, in partnership with HAILab/PUCPR.
Few-shot LearningZero-shotNERTransformers
Jan 2022 – Sep 2022
NLP Engineer
Looqbox · São Paulo, Brazil
Developed natural language parsing and generation modules for a BI analytics tool. Built parsers using ANTLR and Python, RESTful APIs with Flask, and automated deployments via Jenkins and Docker.
NLPANTLRPythonFlask
Apr 2020 – Apr 2022
NLP Researcher
Designed NLP pipelines for Portuguese: spell checkers, named entity recognition (NER), POS tagging, and chatbots. Applied TensorFlow, spaCy, and scikit-learn to healthcare and education datasets.
NERspaCyTensorFlowPortuguese NLP

PROPOR 2026 · Conference Paper
NormaTex-MapSNOMED: Bridging the Gap Between Brazilian Portuguese Clinical Narratives and SNOMED CT
Araujo, I.; Moro, C.; Martinez, L.
Proceedings of the 17th International Conference on Computational Processing of Portuguese (PROPOR 2026), Salvador, Brazil. pp. 1085–1091. Proposes a method for mapping clinical terms in Brazilian Portuguese to SNOMED CT using large language models, combining lexical, contextual, and semantic evaluation strategies.
IEEE CBMS 2025 · Conference Paper
Abbreviation Identification and Disambiguation in Real-World Clinical Narratives
Joaquim, J. P.; Fontes de Araújo, I.; Moro, C.
2025 IEEE 38th International Symposium on Computer-Based Medical Systems (CBMS), Madrid. pp. 21–24. Addresses the challenge of clinical abbreviations in Portuguese medical text using LLMs.
CBIS 2024 · Conference Paper
Avaliação de Grandes Modelos de Linguagem na Extração de Informações Clínicas
Mello, C. E. R.; Schneider, E. T. R.; Silva e Oliveira, L. E.; do Nascimento, J. N.; Gumie, Y. B.; Araújo, I. F.; Moro, C.
XX Congresso Brasileiro de Informática em Saúde, Belo Horizonte, 2024. v. 16. Evaluates LLM performance on clinical information extraction in Brazilian Portuguese.
Research Focus
Current Interests
Clinical text normalization · Abbreviation disambiguation · LLM evaluation for healthcare · Multilingual NLP (Portuguese/English) · Prompt engineering for low-resource settings · Information extraction from biomedical corpora.

2022 – 2023
Postgraduate - Data Science
2014 – 2019
Bachelor of Engineering — Computer Engineering

🇫🇷
BRAFITEC — Brazil–France Engineering Exchange
Télécom Physique Strasbourg · Strasbourg, France · Sep 2018–May 2019
CAPES Scholarship

Selected as a BRAFITEC (Brasil–France Ingénieurs Technologie) scholar — a CAPES-funded program that places Brazilian engineering students in French Grandes Écoles for a full academic year with dual credit. I studied in the Technologies de l'Information pour la Santé track, covering medical image processing, signal processing, computer vision, biomechanics, human–machine interface, optimization algorithms, and advanced statistics.

Exchange ProgramHealth Informatics Medical Image ProcessingSignal Processing Computer VisionCAPES
🇩🇪
Research Internship — Psychology Institute
Johannes Gutenberg University Mainz · Germany · May – Aug 2019
Research Intern

Worked at the Experimental Psychology Laboratory studying pedestrian behavior in traffic scenarios involving electric and conventional vehicles. Collected and processed behavioral data using MATLAB and audio/video recording equipment. Contributed to experimental design, quantitative data analysis, and visualization for human–vehicle interaction research.

MATLABBehavioral Research Human–Vehicle InteractionData Analysis Experimental Design
🇫🇷
Projet Ingénieur — PROTiP Medical (HiPEAC)
Télécom Physique Strasbourg · Strasbourg, France · Sep 2018–May 2019
Undergraduate Thesis

Undergraduate thesis project in collaboration with PROTiP Medical through the HiPEAC research network, focused on patient-specific 3D-printed prosthetic implants. Developed a Python-based automation pipeline to streamline implant sizing from medical imaging data, integrating CAD software and 3D modeling to reduce manual steps in the production of personalized orthopedic devices.

PythonCAD3D Modeling Medical DevicesHiPEACAutomation

NLP & AI

Large Language Models Prompt Engineering NER & Text Classification Clinical NLP Few/Zero-shot Learning

ML Engineering

Python TensorFlow · PyTorch LangChain Hugging Face Transformers spaCy · scikit-learn

Infrastructure

Docker · Kubeflow Databricks Flask · FastAPI SQL MLOps / AI Engineering
Portuguese Native
English Bilingual
French Full Professional
German Limited Working
Spanish Elementary

I'm open to research collaborations on NLP and clinical AI, interesting ML engineering challenges, and conversations about LLMs in healthcare. Feel free to reach out.