Character Name AJ Caylor

Redox Inc.

Senior Software Engineer — Platform

02/2022 – Now · Remote

• Operate AWS and GCP Kubernetes infrastructure supporting 83 production applications, 4,000+ pods, and an EKS cluster that scales between 100 and 200 nodes.
• Managed successive production EKS upgrades from Kubernetes 1.22 through 1.34 as part of ongoing platform lifecycle ownership.
• Designed and implemented a GitOps continuous-delivery platform using ArgoCD and Crossplane, reducing deployment time by 80%.
• Design application-specific scaling strategies using KEDA, HPA, and VPA according to workload behavior and capacity requirements.
• Lead development of a custom canary deployment and automated rollback workflow for the company’s largest application, which operates approximately 1,800 pods.
• Author reusable Terraform modules for AWS and GCP, validating mirrored infrastructure changes in staging before production.
• Built developer-facing CLIs to promote changes into staging, roll back applications during incidents, and manage multiple EKS and GKE environments.
• Led a centralized Prometheus, Grafana, and OpenTelemetry platform with proactive alerting and 20+ multi-cloud dashboards.
• Serve as primary on-call engineer and incident commander, converting production incidents into reliability improvements.
• Authored 15+ architecture proposals and led cross-team initiatives across reliability, delivery, and observability.
• Built LLM-assisted automation for dependency management and platform maintenance, reducing recurring operational toil.

Georgia Tech Research Institute

Research Faculty — Cloud Engineer

09/2021 – 02/2022 · Atlanta, GA

• Built standardized Kubernetes application platforms supporting research workloads and production services across AWS and Azure.
• Designed and deployed a highly available on-premises Kubernetes cluster integrated with enterprise identity systems and VMware infrastructure.
• Led development of a scalable remote-learning lab platform used by 20+ students per class across three curricula.

Georgia Tech Research Institute

Research Faculty — Systems Administrator

06/2019 – 09/2021 · Atlanta, GA

• Automated large-scale bare-metal and VM deployments using Foreman, Red Hat Enterprise Linux, Python, Packer, and Ansible.
• Implemented centralized log aggregation and observability tooling, reducing log-query time from five minutes to under 30 seconds.
• Led migration of critical file storage to a distributed SAN architecture, improving uptime to 99.99%.
• Developed disaster-recovery and business-continuity strategies that reduced MTTR to under one day.

Georgia Tech Research Institute

IT Intern

05/2017 – 05/2019 · Atlanta, GA

• Managed VMware vSphere storage, networking, and hypervisor infrastructure across three data centers with 1 PB of highly available SAN storage.
• Automated VM configuration, software deployment, and upgrade workflows across Linux and Windows environments.

Multi-Cloud Elasticsearch Platform

100 TB+ indexed data · 2,000+ application clients

• Led end-to-end design of a Kubernetes-native ECK platform providing a repeatable, cloud-agnostic data service across AWS, GCP, and Azure.
• Authored requirements and architecture documentation, evaluated alternatives, and built prototypes before production rollout.
• Architected a highly available cluster supporting 100 TB+ of indexed data and serving more than 2,000 application clients.
• Improved query latency by approximately 20% and throughput by 10% while standardizing monitoring and alerting with Prometheus and Terraform.

High-Scale Metrics & Observability Platform

42M+ active metric series

• Re-architected a single-node metrics pipeline as a distributed Kubernetes platform using InfluxDB, Telegraf, and Grafana Mimir.
• Built a Helm-based deployment with S3-backed storage, replacing a custom cloud-init workflow and enabling consistent installations across environments.
• Scaled the platform to 42+ million active metric series in InfluxDB and 12 million series in Grafana Mimir.
• Reduced InfluxDB cold-boot time from approximately 40 minutes to under five minutes and enabled zero-downtime patching for Mimir.