Daniel Navarro Fernández
Junior Backend Developer
daninf77@hotmail.com · Barcelona · Open to remote · github.com/Danielnf777
Summary
Junior Backend Developer trained at 42 Barcelona, where I built foundations from scratch: memory management in C, POSIX signal protocols, custom sorting algorithms, and direct framebuffer graphics — all verified by peer review and automated graders with no partial credit. Currently building production and SaaS systems with Node.js, Next.js, PostgreSQL, and AI integrations.
Projects
Bilingual editorial platform for Real Madrid coverage
- Streaming video proxy with HTTP Range (206) support and SSRF hardening — replaces Twitter widget with zero external JS, keeping INP/CLS/LCP clean.
- LRU cache (2,000 entries) with request deduplication via shared Promise and mutation safety via structuredClone.
- Dual-provider AI pipeline (Anthropic primary, OpenAI fallback) — batched generation, cost tracking, and a 7-step JSON recovery parser for production-grade resilience.
- 15-route Nginx + Express SSR: 6 controllers, full JSON-LD, hreflang ES/EN, 8-sitemap index, programmatic noindex enforcement.
- 4-tier rate limiting keyed by endpoint type (IP for auth, IP+URL for video streaming).
B2B SaaS platform for AI-powered QR menus for restaurants
- End-to-end WhatsApp editing pipeline: plain-text message → GPT-4o parses into structured actions → fuzzy dish matching (Dice coefficient + accent normalization) → confirmation flow → atomic Prisma transaction with full immutable audit log.
- Human-in-the-loop safety: AI only proposes a batch of actions; restaurant confirms before any data is mutated. Idempotency enforced via unique index on WhatsApp message IDs.
- Admin content pipeline: HTML import (3-strategy parser), draft review UI, per-dish AI image generation (DALL·E), and one-click publish to production DB.
- 25+ React components across 5 menu layouts: à la carte, daily menu, cocktails, wines, and compact — with category navigation, dish modals, image zoom, and language selector.
42 Barcelona — Systems Programming
Client-server communication built entirely on UNIX signals (SIGUSR1 / SIGUSR2). No sockets, no pipes. Each bit transmitted as a signal; sigaction + SA_SIGINFO provides sender PID for reliable ACK routing. A 32-bit length prefix allows the server to allocate an exact buffer before reconstruction begins.
Sorts arbitrary integers using two stacks and 12 allowed operations, minimising total operation count. Three-path strategy: hardcoded optimal sequences for n ≤ 3, manual extraction for n ≤ 5, and binary radix sort by index bits for large inputs.
Renders elevation maps as isometric wireframes via direct framebuffer writes. paint_pixel() computes pixel offsets through pointer arithmetic into MiniLibX's raw pixel buffer. Bresenham's line algorithm draws edges with bounds checking on a 2000×2000 canvas.
Skills
- Systems
- C, POSIX signals (sigaction / SA_SIGINFO), manual memory management, radix sort, linked lists, Bresenham's line, isometric 3D projection
- Backend
- Node.js, Express, Next.js 16 (App Router), REST API design, HTTP streaming (Range / 206), LRU cache + request deduplication, 4-tier rate limiting, SSR, PostgreSQL, MongoDB, Prisma ORM, Mongoose
- Frontend
- React 19, React Router 7, TypeScript, Vite, Tailwind CSS v4, Zod, bilingual SPA (ES/EN)
- DevOps
- Docker, Docker Compose, Nginx (reverse proxy + SSR routing), Cloudflare (CDN + origin certs), Linux VPS, Helmet.js + CSP, HSTS
- AI & Integrations
- Anthropic Claude SDK, OpenAI SDK (GPT-4o, GPT-4o-mini, Whisper, DALL·E), Meta WhatsApp Cloud API, batched generation, robust JSON recovery, HMAC-SHA256 webhook verification
- SEO
- 8-sitemap architecture, hreflang (ES/EN/x-default), JSON-LD structured data, programmatic noindex, Googlebot log monitoring
Education
Computer Science — Common Core
Peer-to-peer, project-based. No teachers, no lectures. All work evaluated by peer engineers and automated graders — no partial credit, no hints.
Availability
Open to full-time roles and contract work — remote or Barcelona. MadridFlow source is private (live production system), but happy to walk through architecture, code patterns, or engineering decisions on a call.