Biomedical scientist turned clinical-genomics software developer — 18+ years in clinical laboratory diagnostics, postgraduate training in Medical Genetics & Genomics, now building bioinformatics and health software. 📍 Goiânia, Brazil · 🔗 LinkedIn · 🧬 ORCID 0000-0002-9281-5818

What makes the combination useful: I know how a clinical laboratory actually works — what a variant report must say, what a reviewer needs to trust it — and I build software that respects those requirements.

A web-based clinical genomics platform: VCF → QC → Ensembl VEP → ClinVar / gnomAD / PubMed → in-silico predictors → ACMG/AMP classification with source-verified citations → PDF report. Research Use Only; no real patient data. Production code is private; architecture and design decisions are public → gencodebio-showcase.

• Evidence integrity by design: a classification without a valid, verifiable citation is discarded and never stored.
• Built to be audited: RBAC, TOTP MFA, revocable server-side sessions, audit trail, rate limiting, security headers; documented risk register (ISO 14971 style), requirements traceability matrix and a preliminary ANVISA SaMD technical dossier.
• Engineered for scale: streaming VCF parsing, batched asynchronous processing for exome-sized inputs, signed-URL uploads.
• International from day one: UI in English, Portuguese, Spanish and French; stable API error codes; CI that fails if a translation is missing.
• Zero-cost local profile (built and tested; hosted-instance upgrade pending): ports and adapters for queue, storage, e-mail and LLM; one docker compose up runs the whole stack on a laptop.
• Local reference data (built and tested): Ensembl VEP official cache as a local service (equivalence with the REST API checked on 330/330 variants), a versioned ClinVar snapshot matched by allele instead of rsID, and region-sliced gnomAD v4.1 for 33 panel genes — with a coverage rule so a position outside the sliced regions is an explicit failure, never a false "absent from gnomAD".
• Provenance and reinterpretation: every PDF report records software image, database versions and model/prompt hash; a read-only endpoint flags classifications that may be stale against the current snapshot.
• Command-line client (gencodebio): MFA sign-in, upload, pipeline stages, variants, PDF report, provenance — passwords never on the command line, session revoked on logout, public API only.
• Quality gates: 585 backend (also run against a real PostgreSQL in CI), 42 frontend and 26 CLI automated tests; dependency auditing, SBOM and container vulnerability scan in CI; OWASP ZAP baseline scan.
• Roadmap: reinterpretation screen in the UI, FASTQ→VCF workflows (Nextflow), trio/inheritance, somatic oncology, GA4GH / HL7 FHIR interoperability.

• clinical-genetics-emr — Electronic medical record for genetic counseling with pedigree construction from structured anamnesis (FastAPI, SQLAlchemy/Alembic, 24 entities, 98 unit/integration + 15 Playwright e2e tests, audit logging, role-based access).
• bioinfo-clinical-training — Documented, hands-on clinical NGS learning path on public reference data (NA12878 / GIAB, GRCh38): QC with FastQC / fastp / MultiQC, alignment with BWA-MEM / SAMtools (in progress), with the reasoning and trade-offs behind each decision.

Languages: Python · TypeScript · SQL · Linux (WSL2) Backend: FastAPI · SQLAlchemy · Alembic · PostgreSQL · pytest Frontend: React · Vite · Tailwind CSS · Vitest · Playwright Bioinformatics: FastQC · fastp · MultiQC · BWA · SAMtools · BCFtools · GATK4 · IGV · Ensembl VEP · VCF/HGVS/ACMG-AMP Platform & delivery: Docker · Docker Compose · GitHub Actions (SBOM, vulnerability scanning) · Google Cloud Run · Cloud Tasks · Secret Manager · Cloudflare Learning: Python · machine learning · clinical data standards

• 18+ years in clinical laboratory diagnostics
• Postgraduate — Medical Genetics & Genomics; Biosafety
• MBA — Data Analytics, BI and Big Data (in progress, expected May 2027)
• Studying — Technology program in Artificial Intelligence and Machine Learning

Open to conversations about clinical bioinformatics, health software and genomics teams. Reach me on LinkedIn.

_{Showcase repositories are source-available under noncommercial terms — see each LICENSE.}