Summary

Site Reliability and Platform Engineer with 9+ years building and running infrastructure where downtime is very visible: live sports streaming with peaks of 20–30M concurrent users across three continents, a banking core kept at 99.99% through a fintech-to-regulated-bank transition, and cloud platforms built from scratch. Deep in AWS and GCP, Kubernetes, Terraform, GitOps and observability.

Experience

1. Senior Site Reliability Engineer

   NBCUniversal · Peacock · New York · remote

   Nov 2022 — present3 yrs 11 mo

   • Lead infrastructure for Peacock's live sports events (e.g. NFL), handling traffic peaks of 20–30M concurrent users.
   • Design and run multi-region deployments across Europe, Africa and the Americas, optimizing latency and replication.
   • Architected the network and security layers that integrate AWS and GCP for secure, high-speed data replication.
   • Run the observability stack (Prometheus, Thanos, Loki, Grafana) that product teams rely on during live events.
   AWSGCPMulti-regionPrometheusThanosLokiGrafana

2. Author & maintainer

   Open source · github.com/andresgarcia29

   Oct 2025 — present1 yr

   • ark-cli: one AWS SSO login puts every EKS cluster across all accounts and regions into kubeconfig in seconds.
   • harness-creator, harness-daemon and harness-ui: an agentic engineering harness where agents propose and deterministic gates verify, a Go daemon that observes coding agents live, and its dashboard.
   GoTypeScriptShellClaude CodeMCP

3. DevOps Lead

   Covalto · Mexico City · remote

   Nov 2021 — Nov 20221 yr

   • Led the technical migration during the acquisition of Banco Finterra, taking a fintech to a fully regulated national bank.
   • Designed the architecture and documentation to meet CNBV banking security and data sovereignty standards.
   • Standardized deployments with GitOps and built health checkers in Go and Python, keeping critical banking cores at 99.99% availability.
   GitOpsGoPythonCNBV

4. Senior DevOps Engineer

   GoExpedi · Houston, TX · remote

   Aug 2020 — Nov 20211 yr 3 mo

   • Built the cloud platform from scratch: from basic Docker deployments to a multi-account AWS strategy managed with Terraform.
   • Secured B2B e-commerce platforms for global enterprise clients.
   • Broke legacy monoliths into microservices and established a high-performance CI/CD culture.
   AWSTerraformDockerMicroservicesCI/CD

5. DevOps Engineer

   Rever · San Francisco, CA · remote

   Dec 2018 — Aug 20201 yr 8 mo

   • Built the company's Data Warehouse and Data Lake from scratch, enabling real-time analytics.
   • Implemented GDPR standards for data privacy and security.
   • Helped scale engineering from 2 to 30+ engineers while keeping DevOps best practices.
   AWSData lakeGDPR

Selected work & open source

ark-cli ↗

v0.13.1Go

One AWS SSO login → every EKS cluster across all your accounts and regions in your kubeconfig, in seconds. Scans in parallel, writes kubeconfig natively, retries AWS throttling with backoff.

52 commits32 releaseslast push

$brew tap andresgarcia29/agm && brew install ark --caskcopy

harness-creator ↗

v0.62.5Shell

Claude Code plugin that turns a multi-repo workspace into an agentic engineering harness. Agents propose; deterministic gates verify.

417 commits88 releaseslast push

$/plugin marketplace add andresgarcia29/harness-creatorcopy

harness-daemon ↗

v0.60.0Go

Single Go binary that shows what your coding agents are doing, waiting on, deciding and spending, live, across machines over SSH.

131 commits34 releaseslast push

$brew install andresgarcia29/agm/harness && harness initcopy

harness-ui ↗

TypeScript

React dashboard for a fleet of coding agents: every session, gate and dollar at a glance.

72 commitslast push

Activity · live from GitHub

How I work

1. Boring on purpose. The 3 a.m. version of me has to understand it.
2. If a script can check it, a script checks it. People and models bring judgment, not diffs.
3. Observability before scale. You can't fix what you can't see while 25 million people are watching.
4. Everything in git. Infra, policy, runbooks and decisions. The audit trail is a feature.
5. Small blast radius. Canary first, roll back first, diagnose second.
6. Write it down. Documentation is the on-call engineer you don't have to page.