Rafael COELHO

Senior AI Engineer · MLOps & Agentic Systems

Production AI systems on Kubernetes — from real-time incremental pipelines to adaptive agentic RAG.

Six years engineering production ML systems — Kafka, Spark, end-to-end GitOps. Architect of COELHO Nexus on Docker · Kubernetes · Helm · Terraform · Terragrunt — adaptive Agentic RAG with a multi-provider Thompson-sampling LLM rotator, two-stage retrieval, GraphRAG via Neo4j, and full observability via LangFuse · OpenTelemetry · Alloy · LGTM.

Brazil Available · Senior AI / MLOps roles
Rafael Coelho

By the numbers

6+ yrs
production ML and Data Science engineering
9
public AI / ML projects shipped end-to-end
BSc
Mathematics · Federal University of Paraná (UFPR)

Selected work

All work →

Recent writing

All writing →

Enterprise MLOps platform — production Kubernetes infrastructure with full CI/CD, GitOps, and workflow orchestration

DEC 3, 2025

Production-grade MLOps platform on Kubernetes — Terraform, k3d, GitLab, ArgoCD, Apache Airflow, MinIO, Rancher — with end-to-end IaC, GitOps continuous delivery, and DAG-orchestrated ML pipelines. Validated against a Real-Time ML project.

#kubernetes #mlops #terraform #argocd

Building a local Kubernetes cluster and provisioning DevOps services with Terraform

NOV 10, 2025

How to build a local k3d cluster that hosts GitLab, Rancher, ArgoCD, and LocalStack — fully provisioned with Terraform for cross-OS reproducibility, with a complete CI/CD flow disposable in one terraform destroy.

#kubernetes #k3d #terraform #helm