I’m historically not one for regular updates to my personal site. I focus my time on scalable cloud architecture and development, which leaves time for family and not much else. Site updates and self promotion usually take a back seat. It’s time for an update, and maybe a little self promotion.
My primary focus in recent years has been poured into RackNet, a cloud-native infrastructure monitoring service I created and actively run. Built natively on Google Cloud Platform (GCP), RackNet handles high-frequency telemetry sent via LoRaWAN, and real-time alert delivery.
Building a SaaS product means constantly solving for efficiency, stability, and scale. As RackNet continues to evolve, the natural next step is leveraging AI to build smarter, event-driven features that both create value for customers and optimize site efficiency. This internal push sparked a broader realization: this path I am going down adding AI technology our own businesses are directly applicable to other businesses looking to navigate their own AI transformations safely.
As a consultant for much of my career, I have gone down the certification path many times. However, my plans for taking the certification test have always been sidelined due to pending deadlines and customer feature requests. I am now asking myself once again: Should I obtain certification? Do certifications still matter?
I’m still reeling from everything announced at Google I/O last month (May, 2026). I have been a user of the Google Skills learning platform for quite a while, as well as Pluralsight. I have previously kept my profile private, but now that a bit more of my time is being spent learning new tech, especially AI, I figured I might as well go public and I bring myself up to date with additional technologies.
The Primary Anchor: Google Cloud & Gemini Enterprise
Because RackNet is rooted in GCP, along with all my cloud based projects, and I bring a 15-year tenure as a Google Workspace Reseller, focusing on Google’s enterprise AI infrastructure is the obvious choice. The Gemini Enterprise Agent Platform provides unique capabilities that match production-level business constraints:
- Native Identity & Governance: It maps seamlessly to pre-existing Workspace organizational units and security parameters, keeping corporate data strictly isolated.
- Frictionless Data Grounding: It hooks directly into enterprise data stores (like Google Drive or BigQuery) to build secure Retrieval-Augmented Generation (RAG) workflows.
While the Google ecosystem is my operational anchor, I have of course used most of the other big AI services as well. Claude Code has been a primary tool. For a while I bounced between Claude Code and Gemini CLI, even pitting them againts each other to migrate a MongoDB Go driver from v1 to v2. (Claude was faster and required less input from me, but I liked the Gemini implementation more.)
Since deciding to focus as much as I can on Google offerings, I primarily used Gemini CLI, and now Antigravity CLI.
And then there are local LLMs.
The Hybrid Approach: Why Local LLMs Matter
True enterprise architecture requires understanding the economic and operational trade-offs of the cloud. Relying solely on cloud APIs can expose a business to variable token transaction costs, and not everyone is comfortable sending their internal data to a cloud service even with the substantial level of Gemini Enterprise security.
It’s only prudent to also focus on local open-weight deployments (such as Hermes with Gemma via llama.cpp). Architecting a hybrid stack that utilizes local LLMs offers massive structural benefits:
- Zero Token Transaction Costs: Ideal for handling highly repetitive, lower-complexity local extraction tasks or building initial rapid prototypes without running up a massive cloud bill.
- Absolute Data Sovereignty: Local models allow us to process or pre-scrub sensitive information and PII entirely offline before any data is tokenized and sent to an external cloud network.
- Deterministic Offline Utilities: Ensuring core analytical utilities remain functional even during network or API infrastructure degradations.
It’s also fun. I target specific hardware in software builds all the time during embedded development. I haven’t done that for a desktop or server environemnt in quite a while.
What’s Next: Merging Theory with Velocity
Obtaining modern certifications isn’t the goal, but perhaps a side effect. Gemini Enterprise is useful. The Google ADK is actually fun. Remaining relevant is essential.
Whether I’m optimizing the event-driven microservices inside RackNet, assisting a client deploy vibe-coded dashboards to their organization via Cloud Run and Gemini Enterprise, or helping a local business with SEO, my primary title of "Architect and Developer" is evolving (becoming outdated?) more quickly than I ever expected.
Do certifications still matter?
The line between internal product development and external engineering strategy has always been blurred for me. One feeds the other, and I ingest new technology mroe quickly when the two are aligned. So do certifications still matter, and are there any in my future? Tune in next year for my irregular and most likely long-overdue update to find out, and check for updates to my certifications page.
I am currently open to fractional contract, 1099 consulting, and advisory engagements focused on GCP infrastructure and enterprise agent architecture. Let’s connect.