Hi, I'm **Fernando Landero** · Senior Consultant & Engineer

# I build data, cloud and AI platforms

I design architectures, write production code and automate platforms on **Google Cloud**, **AWS** and **private cloud**. This is where I publish my projects, comparisons and tools.

- **15+ Years** in systems architecture & cloud engineering, the last 7 focused on enterprise data platforms.
- **Multi-Cloud** GCP, AWS and private cloud platforms defined as code and guarded by automated policy.
- **Regulated Data** Banking, energy and retail systems with PII isolation and auditability.
- **1,100+ Models** Production dbt models & 200+ extractors running on one Kubernetes data platform.

## Core expertise

- **Applied AI & MLOps**

  I build agent workflows for clients: multi-agent loan evaluation with LangGraph on GCP, with token cost telemetry, and Gemini agents on Cloud Run with session state.
- **Cloud & Platform Engineering**

  I build cloud foundations as code: multi-project GCP landing zones in Terraform, zero-trust CI/CD and policy checks before a change applies.
- **Data Engineering & Pipelines**

  I operated a platform with 200+ extractors and 1,100+ dbt models for eight business units, with data isolation, backups and cloud cost optimization.
- **Data & Storage Architecture**

  I model data and choose where it lives: Data Vault 2.0 and dimensional models, BigQuery warehouses and open-table lakehouses, decided by schema changes, concurrent writers and recovery.

## Interactive Tools

Try a failure scenario, replay an agent crash or size an evaluation. These tools run in your browser with examples you can change.

- **Failure Lab · Pipeline Simulator**

  Pause a worker or interrupt writes, then inspect the recovery. Compare retry policies against the same simulated incident.
- **Table Files · Compaction Planning**

  Estimate how many files a table keeps from its daily volume, commit rhythm and partitioning, before and after compaction.
- **Agent Replay · Crash Recovery**

  Crash an agent between a write and its checkpoint, then compare four recovery strategies and see which ones apply the write twice.
- **Eval Statistics · Error Bars**

  Put an interval around a pass rate before trusting it, then size a comparison between two variants.

## Contact

If you want to talk about data, cloud or AI, write to me or book a call. I'm in Caracas (UTC-4) and speak English and Spanish.