Software engineer specializing in backend product systems, operational SaaS, and applied AI. Focused on architectural boundaries, reliable execution, and production-grade discipline.

Portfolio · Case Studies · Writing · LinkedIn · Email

🇧🇷 Procurando a versão em Português?

Clique para expandir

Eu projeto e implemento sistemas de backend, plataformas SaaS e soluções de IA aplicada com limites arquiteturais claros, execução confiável e rigor de produção. Trabalho com foco em contratos de API rígidos, validação de esquemas e comportamento pós-lançamento previsível.

I design systems that remain explainable and resilient under real-world constraints—such as network timeouts, rate limits, billing failures, and non-deterministic LLM outputs. My approach favors strong schema contracts, idempotent operations, and thorough runbook documentation over hype-driven engineering.

Multi-tenant customer onboarding SaaS featuring schema isolation, queue-backed automation jobs, and stripe billing.
Next.js · TypeScript · PostgreSQL · Prisma · Stripe

git clone https://github.com/SavioCodes/OnboardPulse.git
cd OnboardPulse
npm install
# Set up DATABASE_URL in .env
npx prisma migrate dev
npm run dev

Lightweight email risk validation API designed around strict OpenAPI contract discipline.
Node.js · TypeScript · OpenAPI · Docker

git clone https://github.com/SavioCodes/MailSieve.git
cd MailSieve
npm install
npm run test

Educational implementation of a neural network from scratch using raw NumPy matrix math.
Python · NumPy

git clone https://github.com/SavioCodes/rede-neural-do-zero.git
cd rede-neural-do-zero
python -m unittest discover tests/

Detailed technical reports and workflows are published at saviofilho.dev/work.

• VOWGRID — Trust and guardrail layer for LLM agents utilizing policy dry-runs, authorization gates, and rollback systems.
• AcessoQR — Mobile auditing platform with weighted scoring, content moderation, and physical QR certification.
• ORCEI — Mobile quote builder with native WhatsApp delivery and inline business AI assistants.

• TypeScript & Node.js: For unified type safety from API contracts down to the database client.
• PostgreSQL & Prisma: To enforce relational integrity and strict database isolation boundaries for multi-tenant SaaS.
• Docker: To guarantee containerized runtime consistency across local development and production.
• Python: For deterministic mathematical evaluations, scripting, and AI integrations.

I write about real-world software constraints and systems design.

• Simulation before execution — Building sandboxes for LLM agents.
• Billing webhooks need replay discipline — Resilient Stripe webhook handlers.
• Contracts beat clever APIs — Choosing schema validation over brittle code tricks.

• Explainability Over Cleverness: If a system fails in production at 3 AM, it should be simple to debug. I write code that is easy to reason about and trace.
• Operational Handoff: Code is only half the job. Complete systems require migration scripts, trade-off documentations (ADRs), and clear disaster recovery procedures.
• Production over Demos: I prioritize well-tested, bounded behaviors over complex features that cannot survive real-world edge cases.