Field epidemiologist (EpiSUS/MS) · PhD · I build health data systems that run in production.
I work where public health meets engineering. 15+ years inside a state health department in the Brazilian Amazon, turning messy surveillance data into tools that people actually use during outbreaks — nowcasting, infodemic surveillance, LLM pipelines.
nowcasting-sus · Python · MIT
Reports from the Brazilian notification system are always incomplete for the most recent
weeks. This package corrects that delay with Bayesian models (PyMC) so decision-makers stop
reading a falling curve that is actually just late data entry.
pip install nowcasting-sus · packaging · CLI · test suite · CI/CD · versioned releases
infodemia-svi · Python · MIT Health misinformation surveillance, running on schedule inside a state health department: collects public news mentions, detects circulation spikes with z-score against a rolling baseline, classifies rumors, and ships a daily executive report. Built with zero external dependencies (Python stdlib only) so it runs inside a locked-down public-sector environment. 48 unit tests, watchdog alerts on signal change.
mba_fakenews · Python / BERTimbau Two-stage triage for health misinformation: a BERTimbau relevance classifier (AUC-ROC 0.758 on holdout, length-bias mitigation applied) separates state-relevant content from noise before a human reviews it, then a veracity model ranks what deserves checking (CRITICAL / RELEVANT / SUSPICIOUS / ROUTINE).
TelegramPdfReporter · Python A Telegram bot for hospital bed-occupancy reporting: it receives the incoming message, parses it, generates the official PDF report and emails it to decision-makers. Unstructured input, standard output, no manual transcription.
More
- animais-peconotentos-brasil-2007-2023 — time-series analysis of venomous animal injuries in Brazil (Sinan), supporting material for a manuscript review.
- curso_r_basico — course material for applied epidemiology in R; taught to health teams and FETP trainees.
- artigo_RESS_malaria · Doutorado — reproducibility repositories for peer-reviewed papers on malaria forecasting and time-series modeling.
Field epidemiology (EpiSUS, Brazil's FETP) — outbreak investigation on the ground, not from a desk. Health information systems: Sinan, SIM, SIH, Sivep-Gripe. Health intelligence for a state with 22 municipalities.
Python (pandas, PyMC, pytest, packaging) · R (time series, spatial analysis, Shiny) ·
SQL · Docker · Git / GitHub Actions · LLM APIs, RAG, vector search · Quarto
Learning next: cloud deployment (AWS/GCP), FastAPI services, orchestration.
22 peer-reviewed papers and 5 book chapters in epidemiology and public health. ORCID
LinkedIn · mvmlima@hotmail.com · Rio Branco, Brazil (UTC-5)