At Instituto Todos pela Saúde, I designed and manage an on-prem AI environment spanning eight servers, six NVIDIA DGX Spark units, and 50+ virtual machines. It runs AI training workloads about 80% cheaper than an equivalent cloud setup and reduced monthly cloud spending by 26% through rightsizing and removing idle resources.
I also build the institute's data platform, including Python ingestion pipelines that process 60,000+ health records weekly and Argos, an LLM-assisted systematic-review screening tool that processed 8,000+ papers in its first seven months. I introduced GitHub Actions CI/CD on a pilot service, helping deploy frequency grow from one to 10+ deployments per week.
Previously at Criação .cc, I automated provisioning and deployments with shell, Docker, and CI/CD pipelines while tuning caching and CDNs to keep high-traffic client platforms above 90 Google PageSpeed scores. Earlier, I managed Linux fleets, Docker, and shell automation at Aztec Online Solutions.
Alongside infrastructure work, I teach software engineering and technical courses at Univates, mentor IT apprentices, and completed an M.Sc. in Applied Computing focused on Edge AI. I bring hands-on experience across cloud, on-prem infrastructure, observability, data pipelines, and GPU/LLM serving.
