Jan 2026 — Present
- ›Development and maintenance of reporting systems and data aggregation pipelines in Go, on a heterogeneous persistence stack (PostgreSQL/AlloyDB on Google Cloud, MongoDB, Redis and Firestore/Firebase).
- ›Integrations with the Meta ecosystem (WhatsApp Cloud API, Facebook and Instagram), including OAuth flows, template system and versioning.
- ›Design and implementation of RESTful APIs with emphasis on performance, concurrency and correctness in critical metrics.
GoPostgreSQLMongoDBRedisFirestoreMeta APIsREST APIs
Oct 2024 — Dec 2025
- ›Started in the support team for 3 months, gaining deep knowledge of the platform, its flows, APIs and third-party integrations.
- ›Moved to the dev team as a front-end developer, building flow components and integrating third-party APIs using React, TypeScript, Redux, Redux Saga and Vite.
- ›Participated in the development of an AI agent interface, contributing to the planning and implementation of the UI for autonomous agents.
- ›Attended client meetings regularly, developing a strong understanding of business rules and stakeholder communication.
ReactTypeScriptReduxRedux SagaVite
UNESP — Universidade Estadual Paulista
2021 — Dec 2025Bachelor's degree, Computer Science
Etec Philadelpho Gouvêa Netto
High School — Technical in Mechatronics
Deep Learning for Rhetorical Move Detection
Sociedade Brasileira de Computação · Sep 2025
Fine-tuned a distilled version of BERTimbau on the CorpusDT corpus to identify rhetorical moves in Portuguese scientific abstracts, achieving an F1-score of 0.94 in sentence classification. Highlights the potential of NLP for enhancing the quality of academic writing in Portuguese.
0.94 F193.97% accuracy6 rhetorical classes52 abstracts → 1,409 examples
The problem
Portuguese lacks tools to support scientific writing. Swales's rhetorical moves — context, gap, purpose, method, result and conclusion — are usually structured intuitively, and the research gap is often left implicit. The previous reference classifier for Portuguese, AZPort, reached 72% accuracy with a Naive Bayes approach.
The constraint
CorpusDT — one of the few annotated corpora available in Portuguese — contains just 52 scientific abstracts, too little to fine-tune a Transformer directly.
The approach
I built a data-augmentation pipeline that preserves the rhetorical meaning of each sentence — back-translation (PT→EN→PT), synonym substitution, random insertion and swap, and random deletion as regularization — applied stochastically and biased toward the minority classes. This expanded the 52 abstracts into 1,409 balanced sentence examples (stratified 80/20 split). On that set I fine-tuned distilbert-portuguese-cased, a distilled version of BERTimbau, with a 6-class classification head, class-weighted cross-entropy, mixed precision (FP16), AdamW with a cosine scheduler, and early stopping on validation F1.
The result
The final classifier reached a weighted F1 of 0.94 (0.9403 macro) and 93.97% accuracy on the validation set — the best known result for rhetorical-move detection in Portuguese. Error analysis showed misclassifications concentrate between semantically adjacent moves — for example, 7.7% of Result sentences read as Method — a subtle boundary even for human annotators. The errors follow a logical pattern rather than a random one.
Graphical Interface for Exascale Systems Simulation
ERADSP 2023 · 2023
Contributed to the GUI of iSPD (Iconic Simulator of Parallel and Distributed Systems), a tool for performance analysis of large-scale distributed systems. Led the redesign and reimplementation of the interface in C++ using Qt, focusing on responsive design, event handling and efficient visualization of massive performance metrics.
My skills
ReactViteTypeScriptNext.jsTailwind CSSNode.jsGoC
Recent projects
AnotaAI
Flashcards app inspired by Anki.
React NativeJavascriptJupyter
Operational System Simulator
Interactive interface using ncurses and low-level technologies to manipulate tasks.
Cncursesthreadssemaphors
Speech to Text (Speech2Text)
A project using threads and OpenAI's Whisper model to convert speech to text.
PythonQtThreadsSignals
Swales - Scientific Writing
Web application with a BERT model (ONNX/PyTorch) to help students write scientific articles, providing real-time feedback on structure and coherence.
ReactTypeScriptBERT
Loveflow
Web CRM app focused on romantic relationships, helping users organize interactions with potential partners. Just a joke between friends — not to be taken seriously.
ReactTypeScript