I'm Gabriel de Jesus Silva, a software engineer since 2018. I build web systems and distributed backends in Node.js and TypeScript, from technical discovery to deployment. Today I work mostly on AI products at Sami Saúde and Untimeless AI.
A meeting intelligence platform. A browser extension captures live captions from Google Meet, Teams and Zoom, and the platform turns that material into a knowledge base for AI chats and agents.
Browser extension that scrapes live captions from Meet, Teams and Zoom.
Conversations over meeting content, with the full history as context.
A first-party MCP server that plugs meetings into any AI tool the team uses.
Connects to third-party MCP servers to extend actions and data sources.
Workspace-based organisation with member and access management.
I've been building software since 2018, designing and implementing scalable, high-performance systems across healthcare, technology, legal, finance and HR.
I work mainly with Node.js, TypeScript and NestJS in distributed architectures (modular monoliths, microservices and event-driven systems with Kafka, RabbitMQ and SQS), supported by CI/CD, OpenTelemetry, Grafana and automated testing.
In recent years my focus has shifted to applied AI: solutions built on generative AI, RAG and MCP, from technical discovery and architecture modelling through to implementation, deployment and validation with real users.
Production systems, own products and open source libraries.
Meeting intelligence SaaS: caption capture through a browser extension, AI chats, an MCP server and team management. Built end-to-end, from discovery to deployment.
AI support agent combining RAG, internal MCP servers and intelligent agents, integrated with Zendesk and internal services. Raised resolution without human handoff from 19% to 53%, with over 85% positive ratings.
System linking patients to healthcare professionals through care teams, integrated into the hospital's micro front-end portal.
NestJS module that declares GraphQL dataloaders through decorators, solving the N+1 problem with configurable batching and caching.
Library that generates object instances for testing from decorators, with support for relations, arrays and value overrides.
Medical scheduling system covering the full flow, from registration and patient care through to billing and payment.
Web and mobile application for shift scheduling and negotiation among network physicians, with dynamic anamnesis forms.
SaaS Net Promoter Score platform measuring patient satisfaction across partner hospital units.
The technologies I use day to day to design, build and operate systems.
Open to conversations about AI products, distributed systems architecture and engineering work.