Abstract
I build and run multi-tenant SaaS infrastructure. The entire platform team, from the rack to the runtime.
Hands-on platform engineer (SRE · DevOps), engineering lead, and co-founder of a lean multi-tenant SaaS. I design, build and keep production alive, and I have led the teams that do it. Plus self-hosted AI on my own metal.
| 60+production tenants | 50+migrated, zero downtime | 40+self-hosted services | 12+years of homelab |
Capabilities
Platform & Infrastructure
Linux, Proxmox/KVM, Docker, GitOps, identity (OAuth 2.0/OIDC, RBAC), webhooks and APIs, plan/apply tenant onboarding. I build the thing that everything else runs on.
Reliability (SRE)
HA & failover: Patroni, replica sets, Sentinel. Observability, backups & DR, incident response with real postmortems.
AI Infrastructure
Self-hosted LLMs on GPU (llama.cpp), RAG and embeddings, LLMOps. AI that runs on my own metal.
Operations & engineering leadership
Teams of 10+ engineers, a 40+ person delivery network, marketplace P&L at Rappi, a KPI turnaround at Contabilizei. I get the business the systems serve, and the people who run them.
Selected work
architecture, decisions, the occasional 3am incidentMigrating 50+ live tenants across data centers, with zero downtime
Hot-standby replication, an atomic cutover, and how "zero downtime" was proven with packet captures instead of optimism.
Read the case →A one-person platform team
Running 60+ production tenants and hundreds of containers across two data centers, by architecture. Including why not Kubernetes, yet.
Read the case →12 years of homelab: my self-hosted AI stack
Local LLMs on a 24 GB GPU, hybrid RAG, a personal knowledge graph, and an autonomous agent that pentests my own SaaS.
Read the case →A whole AI stack on one used RTX 3090
A ternary 27B model with vision, document search and image generation on one 24 GB card, a VRAM scheduler, and what broke along the way.
Read the case →Experience
| 2024-present | Co-founder & CTO · MotorSync | Platform & infrastructure, end to end, for a multi-tenant SaaS |
| 2025-present | Senior Business Operations Manager · Contabilizei | Five operations structures, KPI-driven (+62% productivity) |
| 2022-2025 | Head of Operations, Marketplace · Rappi | Three roles to Head. Delivery P&L, fulfillment, integrations |
| 2021-2022 | IT Director · AML | Rebuilt the tech org on AWS + Docker. 10+ engineers, 40+ people across four companies |
| 2021 | Technical Project Manager · BairesDev | Four projects for US direct-hire clients |
| 2012-2017 | Project Manager · Mectron / Bradar (Embraer Defense) | Radar and defense contract portfolios |
| 2008-2011 | Industrial automation & data integration | OPC↔Web in Java. The original technical root |
Stack
Platform
- Linux · Proxmox · KVM
- Docker · GitOps
- Traefik · Cloudflare
- GitHub Actions
Data & HA
- PostgreSQL · Patroni
- MongoDB replica sets
- Redis Sentinel
- PgBouncer · ZFS · NVMe
Observability
- Prometheus · Grafana
- VictoriaMetrics
- Fluent Bit
- Blackbox probing
AI infra
- llama.cpp · Qwen
- Embeddings · RAG
- pgvector
- GPU inference
Runtime I operate
- Node.js · NestJS
- Next.js
- Bull queues
Identity & security
- OAuth 2.0 / OIDC · JWT · RBAC
- Zero-trust (Tailscale) · Cloudflare Access
- Cloudflare Tunnel
- Secrets mgmt · least privilege
- Semgrep · Trivy · Grype in CI