Hi, I'm
Marcos.
> Backend Developer
I design and build what runs behind the scenes: robust APIs, clean architecture, well-modeled data and integrations that don't fall over when the real world shows up.
Who's behind the server
I'm a backend developer focused on building the part nobody sees but everybody feels: REST and GraphQL APIs, well-isolated business rules and databases that tell the truth. I work with Node.js, NestJS and PostgreSQL, applying Clean Architecture and DDD so the code stays changeable years after deploy.
I shipped a multi-unit commerce platform to production on my own, from the API to the frontend through CI and deploy, with more than 1,900 automated tests. I enjoy the hard backend problems (idempotency, concurrency, money as decimals, webhooks as the source of truth) and at work I automate processes and integrate AI (Google Gemini, RAG) into an omnichannel SaaS product.
Professional experience
Support & Customer Success Analyst
cVortex
At cVortex, an AI-powered omnichannel SaaS platform, I work at the intersection of customer, product and engineering: combining technical support, customer success and building internal tools that drive efficiency and save time.
- Python automations and internal apps that removed recurring manual platform setup work, saving the team around 20 hours a week
- Integrated the Google Gemini API into an automated support chatbot and evolved an existing RAG application, tuning the context retrieval pipeline for more accurate answers
- Integrate and consume REST and GraphQL APIs, writing SQL queries for product diagnosis and support
- Diagnose and fix problems in complex platform workflows alongside the engineering team
- Direct contact with product and engineering, with a strong grasp of the platform's business rules
Featured projects
Backends I've built, focused on architecture, data consistency and best practices.
Nexio · Restaurant platform (API + Web)
Management platform for a multi-unit restaurant chain (franchises): an API with omnichannel orders (app, web, kiosk, counter, pickup), payments and a loyalty program, built with Clean Architecture and DDD per bounded context, plus a web app with customer area, POS, kiosk and admin.
- Clean Architecture + DDD across 9 bounded contexts (identity, catalog, orders, payments, inventory, promotions, loyalty, AI and audit), with a pure domain and cross-context communication only through ports
- Omnichannel orders across 5 channels (app, web, kiosk, counter and pickup) with server-side pricing (anti-tampering), a state machine and optimistic locking on transitions
- Payments with a signed webhook (HMAC) as source of truth: idempotent and transactional, with a decimal-safe money pipeline via a Value Object over big.js (float never appears in any layer)
- Inventory as an append-only ledger: balance derived from transactions and atomic deduction inside order creation, with no destructive quantity update
- Token-metered AI assistant with Google Gemini: balance debit and usage record written in the same transaction, with a context cap to bound the cost per call
- JWT auth with rotating refresh tokens and reuse detection, argon2 hashing, RBAC scoped per unit, keyset pagination and direct-to-storage upload via signed URL
- Next.js web app with customer area, POS, kiosk and admin, JWT in an httpOnly cookie, PT/EN i18n and continuous deployment on Vercel
- 1,846 unit tests and 97 e2e, CI on GitHub Actions on every push (type-check, lint, tests, build) and a Docker image on Render with managed PostgreSQL on Supabase
zNuvo · Omnichannel support SaaS
Multi-tenant customer support SaaS: each company connects its channels (a chat widget on its website today; WhatsApp and Instagram on the roadmap) and agents answer every conversation in real time from a single dashboard. Polyglot monorepo with NestJS and Go services.
- Hexagonal architecture + DDD with a transactional outbox publishing events to NATS JetStream; Protobuf contracts and internal gRPC between services
- Realtime service in Go over WebSocket (versioned protocol), with single-use signed tickets and closing the sockets of deactivated members
- Multi-tenancy with PostgreSQL Row-Level Security and least-privilege database, Redis and NATS users per service
- Auth via Zitadel (OIDC + PKCE, tokens in memory only), email invitations and self-service signup provisioned by a Go worker with retries and dead letters
- React dashboard with a strict CSP and an isolated Preact widget (Shadow DOM launcher + iframe panel, with per-key frame-ancestors)
- Observability with OpenTelemetry and Grafana, Playwright e2e tests, k6 load tests and CI on GitHub Actions
To-Do App · Full Stack
Full stack task app: a FastAPI REST API with PostgreSQL and SQLAlchemy, a React frontend in TypeScript, all containerized with Docker Compose on WSL2.
Technologies I use
My day-to-day toolbox, from the endpoint to the database, through the infrastructure.
Backend
Architecture & Quality
Security & API
DevOps & Tools
Frontend
Databases
Applied AI
Let's talk
Got a project in mind or want to talk backend? My inbox is always open.