I build AI

systems that run

in production.

• Multi-product SaaS ecosystem across three distinct domains: legal, sales automation and core infrastructure.
• AI SDR — a WhatsApp sales-automation platform with autonomous agent orchestration for lead qualification, conversation routing and external tool calling.
• RAG pipeline with vector search over each client's knowledge base, grounding answers in the real business context and cutting hallucination in sales conversations.
• Multimodal pipeline for audio transcription and image analysis of WhatsApp inbound, feeding the result into conversation context and the agent's qualification decision.
• Prompt engineering and evaluation criteria to track answer quality and catch regressions between agent versions.
• Human Takeover, message debouncing, @lid resolution for Click-to-WhatsApp ads, and automated follow-up to recover abandoned conversations.
• LexPerito — maintaining v1 in production while building v2, focused on architectural modernization.
• PurpleTech Hub — multitenancy, SSO/OIDC, Asaas-integrated billing and cross-product access control.

Stack: Vue.js 3, Next.js, NestJS, Node.js, TypeScript, PHP 8.3, MySQL, PostgreSQL, Redis, Prisma, Docker, Google Gemini, embeddings e banco vetorial, WhatsApp via WAHA

• Technical leadership on the architectural evolution of an AI-driven CRM used by 500+ companies, owning architecture, scalability and quality.
• Architected four autonomous agent profiles on WhatsApp — support, SDR, qualification and sales — each with its own goal and behavior inside the funnel.
• No-code layer where the client themselves defined context, product and tone, with the system assembling agent behavior dynamically per tenant.
• Agents able to mutate application state: the AI read the lead's reply and moved the card across kanban stages with no rep intervention.
• Technical standards, code reviews and mentoring, raising team maturity and delivery predictability.

• Intelligent CRM used by 500+ companies, focused on automation, AI and cutting operational effort.
• Integrated the AI agents into WhatsApp, including the pipeline carrying the conversation to the model and the answer back to the lead in real time.
• Context-assembly layer for agents built from client configuration, with tenant isolation.
• Interfaces that cut users' average task time by 30%.
• 50+ REST APIs following DDD and established design patterns.
• Integrations with WhatsApp, payments and Google APIs, automating 60% of operational flows via N8N and Python.
• Real-time WebSocket communication supporting thousands of messages per day.
• Technical documentation and onboarding for 4+ developers, cutting ramp-up time by 50%.

• Directly led 78 staff across multiple hospital operating departments.
• Process standardization and indicators that cut rework by 35% and operational bottlenecks by 30%.
• Python application that cut consolidation and submission of data to health surveillance from 15 days to under 60 minutes.
• Continuous integration with regulatory bodies and inter-hospital systems (CROSS/SIRESP), with 100% compliance on mandatory submissions.

• Internal Python applications automating collection, processing and analysis of hospital data.
• Integrations with government platforms (CROSS/SIRESP) and other hospitals' systems, keeping data in sync across institutions.
• Strategic dashboards and reports for tracking clinical and administrative indicators.

• Academic, financial and pedagogical systems used by 2,500+ active users.
• Responsive interfaces that cut usability-related support tickets by 30%.
• Secure authentication and LGPD-aligned practices, reducing access incidents.

• Built and maintained internal education systems used by hundreds of users.
• Tested and configured third-party API integrations, reducing integration failures.
• Scripts and automations that cut time spent on operational routines.

Stack: Vue.js, Nuxt.js, Django, FastAPI, PostgreSQL, Redis, Kubernetes, Terraform, OAuth2 e JWT