Full-stack developer focused on mission-critical automation: RPA that has to work in production against systems that actively try to block bots - WAFs, fingerprint detection, hCaptcha, reCAPTCHA.
Main case: architected and delivered in production a system that automates government certificate/record issuance across 15+ Brazilian portals (DETRAN, Receita Federal, Tribunais, Secretarias da Fazenda), each with a different anti-bot stack. Client code is not public - the architecture and engineering decisions are: gov-portal-automation-architecture.
📄 Read: What I learned automating anti-bot-protected forms
- Anti-bot bypass: hCaptcha (incl. HSW proof-of-work), reCAPTCHA v2/v3, image OCR, session-token bypass
- Stealth browsers (Camoufox and derivatives) for fingerprint detection evasion
- Certificate-based auth: automated e-CNPJ/e-CPF digital certificate login (A1)
- Internal API reverse engineering to remove fragile UI dependency
- AI vision as a captcha fallback when pure automation isn't enough
- Resilient orchestration at scale: queues, retry/backoff, concurrency control, anti-OOM, deterministic failure handling, monitoring, alerts
- Web3 / Solana since 2020
- 💰 xgate-challenge - multi-currency financial ledger API: snapshot + event-sourced balances, idempotent transfers/conversions/reversals, 110 tests with an enforced 80% branch-coverage gate. README documents the production trade-offs (Postgres row locking, Redis-backed idempotency, per-account partitioned queues).
- 🛡️ gov-portal-automation-architecture - case study: production RPA architecture against anti-bot systems (captcha chain with fallback, stealth browser config, certificate auth, IP-reputation lessons). No client data.
- 🎼 n8n-orchestrator-pattern - maestro + workers pattern for n8n: central scheduling, shared retry with exponential backoff, error classification, alert deduplication. Extracted from a 12-robot production ecosystem
- 🎰 snowball - gamified Web3 raffle system for Solana token holders: real-time pump.fun monitoring, live WebSocket updates, auditable draws with a verifiable seed (Supabase)
- 12-robot ecosystem + orchestrator (n8n) for a US pool company's operations
- Mobile app with AI (iOS/Android) in production
- aep-residuos - mobile app (Expo/React Native) that classifies solid waste from the camera using on-device computer vision (TensorFlow.js + MobileNet)
- Workana proposal automation - bot that drafts and tracks freelance proposal submissions
- Ticketing platform, AI-powered rendering SaaS feature, and a custom SketchUp plugin (Ruby)
Open to freelance/contract anti-bot & automation projects.
📫 toscanopedroall@gmail.com · Maringá, PR, Brazil


