I'm an infrastructure and DevOps engineer.

Most of my work is around Proxmox, Kubernetes, infrastructure as code, and taking AI services from a developer's machine to a production cluster. On the backend side, I write async Python services with FastAPI, message queues, and PostgreSQL.

I also self-host more things than I probably should.

📍 Natal, RN — Brazil

• On-prem Kubernetes (k3s): NVIDIA GPUs shared via time-slicing, MetalLB, NGINX/HAProxy ingress, Longhorn/NFS storage, Harbor, remote BuildKit builds, a Pulp package cache, and a lot of other services (GitLab Runners, Sentry, Langfuse, MLflow, Keycloak, RabbitMQ…) managed with Helm and ArgoCD.
• Infrastructure as code: GCP/GKE (network, bastion, cluster) with Terraform and Ansible; Proxmox VMs with OpenTofu; Ansible playbooks for OS patching and etcd backups.
• Document and audio pipelines: GPU OCR (PaddleOCR), transcription with faster-whisper and speaker diarization (pyannote), NLP with spaCy, PII anonymization (Presidio), and legal document search on OpenSearch.
• ML platform: MLflow, Jupyter, Label Studio, CVAT, and Open WebUI + Ollama for experiments, labeling, and local LLMs.
• Backend: async FastAPI with RabbitMQ for long-running jobs, Valkey for caching, rate limiting, and WebSocket notifications, and PostgreSQL behind PgBouncer.
• Observability: kube-prometheus-stack, Loki/Alloy, DCGM, and cAdvisor, with my own dashboards and alerts for GPUs, volumes, nodes, and queues.
• CI/CD: pipelines on GitLab CI, GitHub Actions, and Jenkins that lint, type-check, test, scan, build, and push images, then bump the manifests that ArgoCD syncs.

Languages

Backend

Quality & Testing

Containers & Kubernetes

IaC, Cloud & Virtualization

Web Servers, Networking & OS

CI/CD & GitOps

Data & Messaging

Observability

AI, LLMs & Agents

• 🇧🇷 Portuguese: native
• 🇺🇸 English: advanced