Services
Flexible engagements — from a one-off review to an ongoing partnership. It all starts with understanding your context and your goals.
CI/CD pipelines as a product: reusable central templates, keyless authentication and incremental migration with zero downtime.
- Golden pipelines: reusable workflows and composite actions for the whole org
- OIDC/keyless authentication — no static keys in pipelines
- Live migration between pipelines (systems running in parallel, zero downtime)
- Security built in: vulnerability and secret scanning in the flow
Service onboarding: weeks → minutes · critical-service deploy −50% · packaging −80% (fintech at scale, ~70 repos)
Running Kubernetes at scale with Git as the source of truth.
- Multi-cluster (EKS/GKE) with GitOps (ArgoCD) and app-of-apps
- Autoscaling with HPA/KEDA and 100% Spot nodes (cost)
- Service mesh (Istio/mTLS), Vault/External Secrets and per-workload identity
- Reliable rollback: git revert + helm rollback with history
~100 nodes in production (100% Spot) · hundreds of applications on GitOps · ~1.8k pods (orders of magnitude)
Diagnosing and hardening production systems under real load.
- RCA separating code vs infra — without masking the symptom
- SLOs and Synthetic Monitoring as the source of truth
- OpenTelemetry, Prometheus/Grafana and enterprise APM
- Observability FinOps: instrumentation without waste
Critical restart-loop unblocked (cause: event loop, not resources) · ~265 GB of over-allocated RAM reclaimed
Infrastructure 100% as code, multi-cloud, with cost under governance.
- Multi-project, multi-org Terraform (AWS and GCP)
- Ephemeral staging environments (spun up in the morning, destroyed at night)
- Drift detection: code-vs-cloud consistency validated at 100%
- IAM as code and centralized secrets
~50 cloud projects governed by code · ~99% code-vs-cloud consistency · ephemeral staging (FinOps)
The platform that serves AI workloads — built on the chassis you already have.
- Production ML lifecycle: training → predict → evaluate
- Zero-cost LLM serving (Ollama/CPU) with a path to GPU (vLLM)
- RAG with pgvector, embeddings and instrumented retrieval
- Evolving toward mature LLMOps: registry, continuous eval and quality gates
Models in production (recommendation, anti-fraud, clustering) · operational LLM serving and RAG
Support for technical leadership and senior engineer development.
- Platform decisions: build vs buy, stack selection
- Architecture evolution roadmaps and low-risk migrations
- 1:1 sessions for tech leads and growing engineers
- Design docs and ADR reviews
Frequently asked questions
How does a consulting engagement work?
It starts with a 30-minute intro call to understand context and goals. From there I design the proposal: a focused assessment, a fixed-scope project or ongoing support — always with clear deliverables and knowledge transfer to the team.
What engagement formats are available?
Three formats: a one-off assessment (e.g., an architecture or CI/CD pipeline review), a fixed-scope project (e.g., a platform migration or observability rollout) and a monthly retainer (recurring work as the team's senior platform engineer).
Is the work remote?
Yes, 100% remote — for companies in Brazil and abroad (Portuguese and English).
What is platform engineering and why would my company need it?
It's treating infrastructure and the delivery pipeline as a product: reusable pipelines, Kubernetes with GitOps, observability and cost under control. The practical outcome is a team shipping faster with fewer incidents — deploys dropping from hours to minutes and service onboarding from weeks to minutes.
My company wants to use AI. Where do we start?
With the platform, not the model. The thesis: the chassis (Kubernetes, CI/CD, observability, IaC) is what serves AI workloads securely and at controlled cost — the AI pieces (LLM serving, RAG, ML lifecycle) are additive on top. A 30-minute call already maps your starting point.
How do we start?
Book a 30-minute intro call on the site (Calendly) or send your context through the contact form. I answer personally, usually within one business day.
Not sure which format fits?
No problem. In an intro call we understand the challenge and design the right engagement for where your team is.
Start a conversation