Data Scientist & AI Engineer with a statistics background. Applied GenAI, machine learning, production-oriented Python systems. US Army GWOT veteran.
Houston, TX
Joined April 2026
- Tweets377
- Following525
- Followers56
- Likes1.4K
China's AI stock bubble has burst. I expect the U.S. reckoning next.
When it comes, AI and tech will have to compete for capital on roughly the same terms as every other industry. That means tomorrow's dominant companies are probably names we haven't heard yet, and "the next Amazon" will finally be understood as a sales pitch, not a business model.
None of this means the technology matters less. If anything, as near-fully-autonomous businesses become common, enterprise AI deployments will become essential to staying competitive.
Adoption will simply have to be profitable. That's the standard every other industry already meets, and tech executives, and Silicon Valley at large, will learn to adjust.
Aaron Horvitz retweeted
Bill Gates paid for the rights to seven physics lectures from 1964 so that anyone on Earth could watch them for free, and almost nobody has pressed play.
They were given by Richard Feynman.
In November 1964 Feynman gave Cornell's Messenger Lectures. He planned six lectures for a general audience, then added a seventh. The series was called The Character of Physical Law. The BBC filmed it.
A year later he won the Nobel Prize.
The films went around colleges for years, and then disappeared from circulation in the 1980s.
Gates had watched them at Harvard. He later said he "fell in love" with them. He has reportedly told people that if he'd seen them earlier, he might have become a physicist instead of starting a software company.
In 2009 he paid privately for the rights, from the Feynman estate, the BBC and Cornell. He never said how much. Microsoft Research put them online for free and called it Project Tuva, after the tiny republic in Central Asia that Feynman spent his last years trying to visit and never reached.
When the BBC license ran out in 2021, Gates renewed it.
Here's what's inside:
> The Law of Gravitation
> The Relation of Mathematics and Physics
> The Great Conservation Principles
> Symmetry in Physical Law
> The Distinction of Past and Future
> Probability and Uncertainty
> Seeking New Laws
The last one has the line every science teacher quotes. If it disagrees with experiment, it's wrong.
Gates said his 10-year-old son could follow them.
They're still free today on Caltech's Feynman Lectures website.
One of the richest people alive paid to make the best physics teacher of the century free.
The only thing left to pay is about seven hours.
Which would you watch first: gravity, or why time only moves forward?
Aaron Horvitz retweeted
To all my fellow veterans, I love you dysfunctional motherfuckers, but let’s be real.
Half of us can’t get out of bed without sounding like a bag of fucking Legos falling down the stairs.
Our knees are shot, our backs are fucked, and we need three business days to recover from mowing the damn lawn.
But ask us about that 12 mile ruck march in 1998, and suddenly we’re all Olympic athletes again.
Every unit had one dumbass lieutenant, one sergeant who hated happiness, and one private who couldn’t find his own asshole with a flashlight.
And somehow, according to every veteran I’ve ever met, we were all the baddest motherfuckers in the entire United States military.
I’d still go to hell and back with you crazy bastards.
Much love and respect to my brothers and sisters in arms.
What happens when Silicon Valley's business model meets real economic constraints?
I recently read a troubling article by Eric Elliott, a veteran software engineer. He describes losing his business, struggling to find work, and becoming homeless.
His story raises hard questions about tech employment.
For much of the past 20 years, Silicon Valley benefited from cheap capital, extraordinary valuations, and tolerance for operating losses. Raising money and chasing growth became a business culture of its own. Few traditional industries enjoyed anything comparable.
Consider someone starting an oilfield services company or building an independent oil and gas producer. They need equipment, employees, insurance, customers, and working capital while navigating regulations and unpredictable commodity cycles. Cash flow matters from day one. Burn through capital without revenue, and investors aren't lining up to fund another expansion.
Compare that with software startups that can lose money for years while raising funding rounds on the promise of growth and market dominance.
These are fundamentally different economic environments. Silicon Valley has spent too long confusing success in one with mastery of business in general.
The notion that you can simply "tech" your way around every business problem is a fantasy. Technology doesn't eliminate economic constraints or the need for sound judgment.
AI is lowering barriers to entry, threatening software differentiation, and putting pressure on margins. Yet too many young professionals have been encouraged to emulate the celebrity-founder model, chasing valuations instead of developing operational and financial discipline.
I've worked with entrepreneurs in energy and other traditional industries who spent decades managing complex operations, tight margins, capital constraints, and brutal market cycles. Many are every bit as capable as celebrated tech founders. They just never had comparable access to venture funding.
I suspect those experienced operators, and younger professionals willing to learn from them, will have a serious advantage. AI may create more lasting economic value by transforming traditional industries than by fueling another generation of software startups. Domain expertise, customer relationships, and operational experience are harder to replicate than software.
And there's the irony: an industry notorious for ageism may soon need the experience it has spent years undervaluing.
Eric's story reminds us that technological progress has real human costs. We need to prepare people for this transition, not assume everyone can simply learn another tool and start over.
The next generation of successful AI leaders may not be those who know the most about building models, but those who best understand the businesses those models are meant to improve.
medium.com/effortless-progra…
#ArtificialIntelligence #Leadership #Technology #Energy
Aaron Horvitz retweeted
Three years ago, I earned $500k/year in tech.
Today, I’m homeless, living in my car.
28 years of experience. Three technical books. AI engineer since the 1990s.
And I can’t find a job.
My story is a warning about the future of work.
medium.com/effortless-progra…
#AI #FutureOfWork
Aaron Horvitz retweeted
Google just open-sourced the plugin thousands of its engineers use every day to stop AI from shipping sloppy code.
It's called the Elephant-Goldfish Model.
No code until the plan survives. No AI grading its own homework. No context collapse halfway through a feature.
The idea is simple and kind of brilliant.
The Elephant is one long-running AI session that remembers everything: the architecture, the intent, every small decision. It writes zero code. It interviews you, argues with you, and keeps the record.
The Goldfish are fresh AI sessions with no memory at all. They never saw your design conversation, so they don't inherit its blind spots.
Here's how it works:
→ egm-brainstorm: you and the Elephant debate options under a strict no-code rule
→ egm-draft: you produce a Markdown design doc with the problem, the plan, the alternatives, and the exact files to change
→ egm-review: one Goldfish has to explain the design back. If it gets it wrong, the doc is ambiguous and goes back for redrafting
→ Then several Goldfish attack it as hostile reviewers hunting edge cases, hidden dependencies and logic gaps
→ A final Goldfish makes the go/no-go call: are file paths, data structures and function signatures all present and unambiguous?
Here's why this matters:
Google's team says AI made coding faster but also sped up slop, bugs and session context collapse. Speed without guardrails just scales your mistakes.
This flips the order. The design gets attacked before a single line exists.
Here's the wildest part:
This isn't a research demo. Google says thousands of Googlers run it daily to ship production code.
The repo ships the Markdown rules, the reviewer personas and the review workflows. Free, Apache 2.0.
The honest trade-off: it only runs inside Antigravity for now, with other environments coming. And security isn't on the default reviewer list. You'd add your own security Goldfish.
github.com/google/elephant-g…
Aaron Horvitz retweeted
FBI Hostage Rescue Team (HRT).
Founded in 1983, HRT is capable of deploying anywhere in the US within 4 hours, responding to major terrorist incidents, hostage situations, and high-risk criminal threats.
Aaron Horvitz retweeted
A ‘hello world’ milestone for Project Suncatcher 🚀
After launching our prototype satellite a few weeks ago, we’ve now successfully sent a command up to orbit, booted up the TPU, ran the prompt, and received our first complete response back on earth - a haiku!
Aaron Horvitz retweeted
This is Mike Vining, an operator with the Army's SMU, at a Ranger camp in 1985. If the photo looks familiar, that's because it has been shared countless times online. And no, he wasn't a Ranger before he went through selection. The SMU did sometimes train alongside the Rangers, and that's how he ended up in this shot.
His actual background was explosive ordnance disposal. After a combat tour in Vietnam, where his EOD team destroyed the largest enemy ammunition cache found in the war, he joined the newly formed SMU in 1978. He went through the unit's selection course and graduated from its very first Operator Training Course, becoming one of its founding members and its first EOD specialist.
Over the years that followed, Vining took part in Operation Eagle Claw, the assault on Richmond Hill Prison during Urgent Fury, Desert Storm, and Uphold Democracy in Haiti. He served as Chief EOD Technician and Breaching Sergeant Major within the Unit, and later investigated the 1996 Khobar Towers bombing. He retired as a Sergeant Major in 1999.
More than just a meme, he's one of the original operators.
Aaron Horvitz retweeted
For fourteen years, everyone thought U.S. Air Force Technical Sergeant John Chapman died in the first two minutes on top of a freezing 10,000-foot mountain in Afghanistan.
Then modern image technology unlocked the thermal camera footage recorded by a Predator drone circling high in the clouds on March 4, 2002.
When Pentagon investigators watched the enhanced screen, hardened combat generals broke down in tears. Because the video proved that after everyone else had left the mountain peak, John Chapman woke up alone in the snow and fought an entire army by himself.
That night on the peak of Takur Ghar, thirty-six-year-old Chapman, a father of two little girls from Connecticut, had charged uphill through thigh-deep snow alone to silence an enemy bunker and save his pinned-down team. Hit from a second bunker, he fell motionless. Believing he was dead, the surviving team was forced to retreat down the ridge.
Chapman was not dead.
The drone footage showed him regaining consciousness in the pitch-black snow, surrounded on three sides by enemy fighters, with nine separate wounds in his body.
In total darkness, with no comrade beside him and nobody on earth watching except a silent camera in the sky, he refused to surrender. For more than an hour, he fought from inside the bunker, killing an enemy fighter in bare-handed close combat when they tried to rush him.
Then, at dawn, Chapman heard the heavy thump of rotor blades in the mist.
An American Chinook helicopter carrying twenty-three U.S. Army Rangers was flying straight toward the peak to rescue him, right into the sights of an enemy rocket launcher.
Chapman knew that if he stayed safe inside the rock bunker, those twenty-three young Americans in the helicopter would be blown out of the sky.
So on the thermal video, investigators watched John Chapman deliberately climb out of the bunker, stand up tall in the open snow, and fire into the enemy positions to draw every machine gun on the mountain onto his own chest.
He died in the snow as the sun rose, shielding the incoming helicopter with his life.
True bushido is what a warrior does when he thinks nobody will ever know his story. The sky was watching, John. And America will remember your name forever.
🤖 Made with AI
Aaron Horvitz retweeted
The #1 Danger Isn't Roblox
Ryan Montgomery on the #1 thing parents get wrong about protecting kids from online predators.
Check out the FULL VIDEO here: youtu.be/5wZUwZWxMLQ
This is insane! Fireworks just posted this demo. They’re reporting 1,521+ tokens per second with DeepSeek V4.1 Flash on Fireworks Ultrafast. It built a playable Tetris game built in 11 seconds! That's an insane pace for prototyping!
DeepSeek V4.1 Flash 1017 runs at 1,521+ tok/s on Fireworks Ultrafast.
Fast enough to ship Tetris in 11 seconds, 7x faster than Claude Opus 5.5.
Join the Ultrafast waitlist: bit.ly/3UA0wKq
Aaron Horvitz retweeted
This is serious.
I just saw with my very own eyes that even non-technical people can exploit CVEs like PHP2Shell using certain AI models and I am actually frightened.
Once you figure it out, it is downright trivial to do. The digital infrastructure of the world is at risk.
Aaron Horvitz retweeted
🚨 Nightmare day for cybersecurity
Multiple major security incidents, active exploits, supply chain compromises...
If you run production systems, start patching NOW.
PHP: Two critical unauthenticated RCEs. Attackers already exploiting them. Patch NOW
LiteLLM: v1.107.0 on PyPI and Docker is BACKDOORED. Steals cloud, SSH, and AI credentials. Pin 1.106.2. Rotate secrets.
eMudhra: A certificate authority's keys were crackable. Update your browsers, systems, and trust stores.
Node.js / Next.js / React: Security releases for Brotli RCE, TLS bypasses, and TWO critical Next.js RCEs.
TiDB Cloud: PingCAP breached through stolen credentials. Customer information and other sensitive data exfiltrated. Read the attacker's release, fun stuff.
Plex2Shell: Weaponized within HOURS of the PoC. Attackers already deploying cryptominers and proxy backdoors.
BeyondTrust / Ivanti: Actively exploited zero-days added to CISA's KEV catalog. BeyondTrust attacks linked to a China-nexus campaign.
And that's not even the full list.
Patch. Audit. Rotate. NOW.
Aaron Horvitz retweeted
Sound on!
Gemini 4 Argon built a 3D periodic table synced to Tom Lehrer’s song, “The Elements.”
Created in @Antigravity, it animates orbiting electrons, atomic data, and light spectra.
Aaron Horvitz retweeted
25 years ago today, Green Berets entered Afghanistan following 9/11.
We honor those who answered the call, remember those who never returned, and stand with the families who shared in their sacrifice.
Their legacy endures. Our commitment continues.
De Oppresso Liber.
Aaron Horvitz retweeted
🚨 ESTO ES UNA MINA DE ORO
Anthropic ha publicado +400 agentes y skills listos para usar en casi cualquier tarea:
→ github.com/msitarzewski/agen… (158k estrellas)
Unos 170 agentes organizados como una empresa: ingeniería, diseño, marketing, ventas, producto, testing, seguridad y soporte.
Con packs adicionales para finanzas, salud y desarrollo de videojuegos.
Compatible con más de 15 herramientas.
→ github.com/wshobson/agents (40k estrellas)
202 agentes, 184 skills y 105 comandos repartidos en 94 plugins para Claude Code.
Especialistas en lenguajes de programación, revisión de código, depuración, infraestructura, seguridad y testing.
Instala solo los plugins que necesites.
→ github.com/anthropics/knowle… (26k estrellas)
Los plugins del propio Anthropic, organizados por áreas de trabajo: ventas, atención al cliente, gestión de producto, marketing, legal, finanzas y datos.
→ github.com/anthropics/claude… (18k estrellas)
Agentes oficiales que puedes poner en marcha: un resumen diario de Slack y GitHub, clasificación programada de incidencias de Sentry, bots para Slack y Linear, un agente de atención al cliente y un analista de datos financieros.
→ github.com/addyosmani/agent-… (103k estrellas)
De ingenieros de Anthropic.
25 skills que cubren todo el ciclo de desarrollo: interview-me, redacción de especificaciones, desarrollo guiado por tests, pruebas en navegador, revisión de seguridad e integración y despliegue continuos
Yo empezaría por claude-quickstarts.
Cada agente es un ejemplo completo y funcional que puedes ejecutar y desmontar para entender cómo funciona.
Aprovecha estos recursos oficiales gratuitos para tus próximos proyectos👇
Aaron Horvitz retweeted
holy sh*t. Karpathy's second brain pattern runs on Grok Bot now: stop explaining your life to AI from zero every chat, keep a wiki that writes itself
three bots in one group chat, each with its own memory:
the reader takes every link, PDF and voice note you drop and never edits it
the librarian rewrites every wiki page that source touches, usually 10 to 15 of them
the checker wakes up on a nightly routine and hunts contradictions, stale claims and pages nothing links to
→ you ask a question: the answer gets filed back as a page, so the brain grows from use, not only from reading
→ a new source contradicts an old page: the checker tags you instead of quietly overwriting it
→ a page goes 30 days without a fresh source: it gets marked unconfirmed
a second brain can sound sure while remembering the wrong thing
Reader collects. Librarian writes. Checker doubts.
every page has to show:
> the claim
> the source it came from
> the date it was last confirmed
paste this into Grok Bot:
"Build me a second brain with three bots in one group chat. Keep my sources untouched in raw/, maintain wiki/ yourself, file good answers back as pages and run a nightly check for contradictions. Show me the structure first. Do not create anything until I confirm"
start with one topic you actually work on, not your whole life
full setup below
