Production AI infrastructure. From the plan to the operation. 🇧🇷 IA em produção: do plano à operação.
I run WRP Consultoria in Brasília, Brazil. 12+ years of engineering across software, telecom and systems operations.
- Private inference: models on your own servers with vLLM, with memory and request limits
- Shared GPU: text and voice models queued on the same card, with wait time and capacity measured
- AI with sensitive data: server-side permissions, protected storage, human-reviewed outputs
- Voice & telephony: real-time speech-to-text (WebRTC, SIP, Asterisk) that recovers from failures
- Integration: AI wired into existing databases and systems, step by step, without stopping operations
| Product | What it does |
|---|---|
| EditalMD | Brazilian public tenders (PNCP): alerts, tender as Markdown, winning-price insights |
| Radar CNPJ | Look up any Brazilian company across 40+ public datasets: partners, sanctions, debts, contracts |
| PontoFato | Any Brazilian ZIP code: Census 2022 data and nearby businesses |
| Agentalog | MCP servers & Agent Skills catalog, with security checks and change alerts |
| HookPulse | Know when a backup, cron job or automation stops running |
| PageAudit | SEO, security and speed audit with fixes |
Most of these come with an API and an MCP server.
- HookPulse check-in: one step that alerts you when a scheduled workflow stops running or fails. No secret. To watch every scheduled workflow with no YAML, install the HookPulse app.
- PageAudit preview check: audits a deploy preview against production and shows only the SEO issues that got worse. The PageAudit app adds this check to each deployment.
🌐 wendelmaques.com · ✉️ eu@wendelmaques.com · 💬 WhatsApp
