Grace Hopper's Nanosecond: The Physical Lesson We Forgot
She handed out wire segments to teach latency. Today's engineers think the cloud is magic. Physics hasn't changed.

45 Years Building What Others Said Couldn't Work
I toggled my first program into an Altair via front-panel switches in the late 1970s. In the 1980s, I ran BBSs when "online" meant dialing in at 300 baud. In the 1990s, I had two full T1 lines running into my house (3 Mbps total, when most people were on dial-up) because I was running servers before "the cloud" existed.
By the 2000s, I'd upgraded to an 8 Mbps point-to-point microwave link feeding a 60-server data center in my basement, complete with dedicated cooling. I was running "the cloud" before AWS existed.
I spent the 2000s surviving the crash. I spent the 2010s fixing the messes left by "move fast and break things."
I served aboard the USS Missouri during Desert Shield and Desert Storm. The Navy taught me that systems don't care about your feelings. If you ignore physics, ships sink. If you ignore economics, startups die.
I don't sell "digital transformation." I sell reality checks.
My late wife MJ and I donated everything we owned and traveled the world together for over eight years—until COVID forced us to pause. Just our backpacks and each other. We lived in Vietnam, made Tokyo's Shinjuku our favorite haunt, explored South Korea, found the best street vendors in Bangkok—chasing great food wherever we went. We called ourselves the Roaming Pigs.
It was the most rewarding and meaningful time of my life.
I created The Mariajuanita Guzman Art Foundation in her memory, and I started RoamingPigs to help fund it. This is my tribute to her.
The spirit carries forward into how I work: go where the good stuff is, commit fully, never half-measure anything. I don't sit still. I don't stay in the corporate pen. I go where the problem is.
The Constitution I Work By
Latency is determined by the speed of light. Costs are determined by supply and demand. No amount of "Agile" or "Vibe Coding" can negotiate with these realities.
Every line of code you write is a line you have to debug, secure, and upgrade. The best architect is the one who talks you out of building the feature.
I don't have a certificate that says I know how to scale. I have the scar from when my system hit 30 million concurrent connections and didn't crash.
Vendors lie. Dashboards lie. AI hallucinations lie. My job is to be the professional skeptic who checks the chrome://sandbox and reads the assembly.
Beautiful systems that lose money aren't beautiful. Ugly code that prints cash is a masterpiece. I optimize for survival, not elegance.
I've watched the same mistakes play out across the dot-com crash, the blockchain bubble, and now AI. Pattern recognition beats novelty every time.
Systems that taught me what works and what doesn't
Architected a push platform that handled 30 million concurrent connections (simulated via 3,000 AWS instances). It didn't break. Designed to scale to billions.
Deployed speech recognition for the US Coast Guard in environments so noisy that standard ASR failed. This wasn't a demo; lives were on the line.
Acoustic intelligence for environments where standard ASR fails. Maintaining 90-95% accuracy where commercial systems drop to 14-40%. The next scar in progress.
Built content management systems before "CMS" was a term. MSNBC's publishing platform in the mid-90s, when the web was still young and everything was being invented.
Acoustic intelligence ASR. Real-time speech recognition that maintains 90-95% accuracy in high-noise environments where commercial systems fail.
ambie.ai →Voice analytics platform. Extracting actionable intelligence from conversations at scale—speaker identification, sentiment, compliance, and real-time insights.
auralytic.ai →Autonomous AI that acts as a complete family office—managing wealth, optimizing taxes, booking travel, and tracking health—for a fraction of traditional costs.
abundera.ai →Technology consulting that sees it through. Home of the Field Manual—160+ articles on what actually works in tech.
roamingpigs.com →Software studio for startups. Building products that work, not products that demo well.
barbarians.com →Lessons from 45+ years of building systems that work
She handed out wire segments to teach latency. Today's engineers think the cloud is magic. Physics hasn't changed.
The industry solved "big ball of mud" by creating a distributed big ball of mud. The complexity didn't disappear—it just got harder to debug.
Every accuracy claim on that slide deck was measured under perfect conditions you'll never have. Here's how to test what they won't show you.
What I'm working on right now
Acoustic intelligence for high-noise environments. Building an ASR system that maintains 90-95% accuracy where commercial solutions fail at 14-40%.
I am expensive because I am fast, and I am fast because I have seen your problem before. If you want a "Yes Man" to validate your roadmap, hire a Big 4 consultancy. If you want to know why your cloud bill is $50k/month or why your AI agent is hallucinating, call me.
Start the Conversation cisco@roamingpigs.com