Blog

From building networks to learning software and AI infrastructure. These are the lessons, decisions, and homelab examples I want to share with the next person starting out.

Start here

Giving back: a homelab series built from my journey

From networking and Full Sail to platform engineering and AI infrastructure: why I am sharing a practical, connected homelab series.

My journey and the homelab series →Thank you, Full Sail →

Let the Hyperscalers Buy the GPUs

October 5, 2026

Capital, power, depreciation, and physics shape AI infrastructure. My opportunity is turning proprietary knowledge into useful intelligence with open-weight models.

Observability Across Cloud and Datacenter Boundaries

October 3, 2026

A Mimir tenancy milestone and the platform architecture behind CloudSpinUp.

When Kubernetes Is Healthy but the Network Still Matters

October 3, 2026

Bringing network engineering habits into RKE2, Kube-OVN, and reproducible platform infrastructure.

Thank You, Full Sail: Bringing My Engineering Journey Into Computer Science

October 3, 2026

Looking back on completing my B.S. in Computer Science in November 2025, building VoiceRAG, and connecting a career in networks and reliability to AI infrastructure.

Giving back: learning Vault beyond a development server

October 3, 2026

A small Vault lab teaches secret access, policies, and recovery without pretending a tool removes the need for careful operations.

Giving back: what Rancher adds to a Kubernetes lab

October 3, 2026

Cluster management becomes more useful when you understand the clusters, identities, and recovery behind the dashboard.

Giving back: a private resolver you can explain

October 3, 2026

Recursive DNS is a useful place to learn caching, access boundaries, and the difference between authenticity and privacy.

Giving back: authoritative DNS and clear private boundaries

October 3, 2026

A small PowerDNS example shows how owned records differ from recursive lookup—and how to test the answers you intend to expose.

Giving back: observability that answers a real question

October 3, 2026

A small LGTM lab connects metrics, logs, and traces while keeping collection, storage, and coverage limits visible.

Giving back: understanding Kubernetes with kubeadm

October 3, 2026

Building a small cluster makes Kubernetes networking and operating responsibilities easier to see.

Giving back: making GitLab a delivery foundation

October 3, 2026

A local GitLab can make delivery understandable—when its bootstrap, runner permissions, and recovery path are designed first.

Giving back: database changes and honest delivery boundaries

October 3, 2026

Flyway helps make migrations repeatable, while clear repository ownership keeps infrastructure, applications, and data changes understandable.

Giving back: a public service without opening the whole lab

October 3, 2026

A Cloudflare tunnel can provide a public path to a service, while network segmentation supplies a separate boundary.

Giving back: the same infrastructure questions on bare metal

October 3, 2026

Moving a lab onto physical machines makes storage, access, and failure domains harder to hide—and useful to understand.

GitLab as a Delivery Backbone

October 3, 2026

Lessons from separating application delivery from infrastructure changes and making deployment evidence explicit.

AI Infrastructure: Evidence Before Promotion

October 3, 2026

How my network and reliability background shapes model evaluation, artifact provenance, and operational safety.

Month 5 Week 4: Five Month Reflexion - From Idea to Real-Time Communication

November 23, 2025

A retrospective on five months of building an audio-first AI communication system. From design doc to deployed product, navigating VAD, Whisper, LLM architecture, and GPU optimization to achieve real-time, conversational AI.

Month 5 Week 3: GPU Migration and Real-Time Performance Optimization

November 16, 2025

This week focused on moving Whisper from CPU to GPU and splitting LLM inference across dual GPUs. The goal: achieve real-time end-to-end communication with sub-second time-to-first-response metrics.

Month 5 Week 2: GPU Optimization and Launch Preparation

November 9, 2025

This week I tackled GPU hardware expansion and AI pipeline performance optimization. Installed a second GPU, worked through optimization challenges, and focused on bug bashing as we get closer to launch and demo.

Month 5 Week 1: End-to-End Audio Pipeline Milestone

November 2, 2025

Major milestone achieved - receiving audio from the AI pipeline works end-to-end! Further optimization needed for speed, but the foundation is in place with Stasis application orchestrating audio flow through the Kubernetes GPU cluster.

Month 4 Milestone 4: FusionPBX Collaboration and Shared Vision

October 25, 2025

Swapping the Back2Back User Agent led to a meaningful collaboration with FusionPBX's Lead Developer, sharing a common vision of AI augmenting humans and creating memorable experiences rather than displacing them.

Month 4 Milestone 3: GPU Optimization and Scaling Challenges

October 19, 2025

Overcoming GPU cluster challenges with power supply swap, achieving 30 concurrent calls with headroom for more, and addressing CPU cluster scaling needs while working toward E2E audio forking with WebSockets.

Month 4 Milestone 2: E2E Milestone Progress

October 11, 2025

Significant progress toward the end-to-end milestone with ingress working on all clusters, addressing minor analytics issues, and preparing for audio forking and system demonstration.

Month 4 Milestone 1: Cost Optimization and E2E Progress

October 4, 2025

Strategic infrastructure shift to home lab via dual IPSEC tunnels to manage AWS costs, while making progress on forking audio and connecting subclusters toward the E2E milestone.

Month 3 Milestone 4: Four Kubernetes Clusters Deployed

September 28, 2025

Major deployment milestone with four Kubernetes clusters now live: Voicerag App, AI, Audio, and Supabase clusters, setting the stage for inter-cluster communication and end-to-end workflow.

Month 3 Milestone 3: Consolidating Flyway Migrations

September 22, 2025

Battling competing Flyway migrations and consolidating 27 DDL scripts for the Supabase persistency store as we near the end of the contextual engine deployment.

Month 3 Milestone 2: Complexities with Context

September 15, 2025

Navigating database migration complexities between Supabase services, addressing race conditions in DDL scripts, and working through context engine memory issues while maintaining development momentum.

Month 3 Milestone 1: Context Engine and AI Architecture

September 8, 2025

Deep dive into building the context engine and AI portion of voicerag.ai - constructing six services across two Kubernetes clusters for scalable, robust AI-powered voice interactions.

Milestone Month 2: FreeSWITCH Breakthrough

August 22, 2025

Navigating the complexities of FreeSWITCH to achieve working inbound and outbound call functionality for voicerag.ai - a critical milestone in the voice infrastructure development.

Reaching Milestone One: The Journey Behind Voicerag.ai

August 15, 2025

Reflections on reaching Milestone One for voicerag.ai - building middleware API and an AI LLM engine capable of understanding context and semantics in natural human phone conversations, running at the edge for low latency and privacy.

Month 2 - Capstone: No Summer Break When You're Shipping VoiceRag.ai

August 4, 2025

Month 2 of VoiceRag.ai development during capstone - building through summer break with terminal windows, Git commits, and late-night debugging to create middleware that bridges legacy voice infrastructure and modern AI.

VoiceRAG.ai Journey – Month 1

July 24, 2025

Reflections on the first month of VoiceRAG.ai development - from concept to concrete execution, covering pitch decks, design documents, wireframes, and the foundation for an AI-driven voice workflow platform.

Pivoting with Purpose: Introducing Voicerag.ai

July 20, 2025

Introducing Voicerag.ai - middleware that connects modern phone systems to powerful agentic AI, transforming customer interactions through natural conversation instead of rigid IVR menus.

Efit Pre Onboard: My Research Pre-Onboarding on echelonfit.com

July 14, 2025

Research findings on echelonfit.com's infrastructure, technology stack, and business model before onboarding - covering DNS, CDN, email services, analytics, and their comprehensive fitness platform.

My Journey: From Network Engineering to DevOps to Developer – A Story of Continuous Evolution

July 7, 2025

A reflection on my journey from self-taught network engineering through DevOps and cloud infrastructure to becoming a developer, with 15+ certifications and experiences at companies like Magic Leap, Chewy, and Firefox.