“The only limit is the speed at which we learn.”-@sama | @ns C1 | @gauntletai C2 Grad | Prev Ranger Medic | building for @alphaschool
Austin, Texas
Joined August 2009
- Tweets12.7K
- Following1.9K
- Followers6.8K
- Likes42.2K
Pinned Tweet
For those who are interested in working in EdTech, I need to be honest with you about what you're signing up for.
You probably shouldn't apply.
EdTech's graveyard is full of brilliant engineers. AltSchool raised $200M and closed. Knewton raised $180M and sold for scraps. 2,148 edtech startups in India shut down in the past five years.
They didn't fail because of a lack of technical skill. They failed from mission drift.
Here's what working in edtech actually means:
You're not building software. You're building a theory of how children learn. - @MitchForest
When you're 2 months into the same feature, and it still doesn't work, you won't have the dopamine hits of shipping fast. You'll need something deeper.
For the engineers who build learning apps for @AlphaSchoolATX, they are REQUIRED to spend 2 hours every day studying. Not coding. Studying and reading papers on cognitive load theory, motivation research, and translating learning science into applications.
Two hours. Every single day.
Engineers like @yiran__c write about the neuroscience of handwriting. @arpangup shares the Harada Method. @LamarDealMaker explores AI for skill development.
This isn't optional. It's the job.
Your success metrics will lie to you.
High DAUs don't mean kids are learning. Viral engagement doesn't mean mastery. The metrics that drive consumer tech success actively undermine educational outcomes.
The money will tempt you to build the wrong thing.
EdTech market: $400B by 2030. That capital creates pressure to optimize for pitch decks instead of classrooms. Byju's hit $22B before collapsing.
An MBA analyzes constraints. A builder changes reality.
What we require:
➡️ Mission Alignment
➡️ Agency
➡️ Strong Engineering Skills
If you're chasing market opportunity or building for your resume, don't apply.
But if you're a builder who sees broken systems and feels an overwhelming need to fix them? If you believe high standards create happy kids? If you want to change education for a billion children?
Before applying, read the link in the reply.
And if you're still interested after understanding what it really takes, let's talk about joining us in building the future of EdTech.
Patrick Skinner - edu/acc retweeted
I used to be the person companies called after they got hacked.
And I noticed something early on.
The hack almost never sank them. The cover-up did.
Every time, the team’s instinct was the same: wait until we know more, wait until we're sure, wait until we have the perfect statement, etc.
And some of the biggest companies in the world have learned what that waiting costs.
In 2016, Uber got hacked. Instead of telling anyone, it paid the hackers $100K to delete the data and stay quiet, and labeled it a bug bounty. The truth came out a year later. Uber paid $148M to settle with all 50 states, and its former security chief was criminally convicted.
That same year, Yahoo revealed that a 2014 breach had exposed 500M accounts. The news landed in the middle of Verizon buying the company. Verizon cut $350M off the price.
And in 2017, Equifax discovered a breach affecting over 140M Americans, then waited 40 days to tell the public. During that window, an executive who'd figured out what happened sold nearly $1M in stock. He went to prison for insider trading.
Meanwhile, hackers don't wait. They post your data. They email your customers directly. And when customers hear it from a stranger first, that trust is gone for good.
Telling customers early doesn't mean admitting fault. It can be as simple as: "We found suspicious activity. We're investigating. Here's what you should do right now. We'll update you as we learn more."
On Tuesday, I'm sharing the exact framework I'd use to write that email if I were in your shoes. Follow and turn on notifications so it's there when you need it.
Customers will forgive you for getting hacked. Most of them expect it by now.
What they won't forgive is finding out you knew and said nothing.
Patrick Skinner - edu/acc retweeted
Everyone wants to talk about the AI @Alphaschool, and it’s understandable, because the software is impressive and continuously improving…but my daughter went from hitting her learning goal 0 days a week to all 5 a week in a month, and the difference came from her guide. 🧵
Patrick Skinner - edu/acc retweeted
“I have seen the MAP data, so I know Alpha students are legitimately performing well ahead of the pack. How much are the results to date selection effects? Chattering classes want to know…
To me, the better question is: can the world’s best learning scientists create a comprehensive, cohesive curriculum, instruction, and assessment system that can deliver breakthrough outcomes regardless of the student population?
On that question, based on what I’ve observed, I’ve voted with my feet to join Alpha.
My fervent hope—and belief—is that Alpha will soon move from startup to pace setter, at a moment when K-12 education desperately needs one.”
Every time one of the engineers I work with have an opportunity to tour @montessorium, I practically BEG them to ask for one of the guides to give them a lesson.
The lesson could honestly be on anything, I simply want them to experience it.
@miss_hannahmc and @annamorgsmiley are absolutely second to none. They embody what it means to nail lesson delivery every single time.
EdTech engineers should really learn and experience what it means to deliver scripted lessons with stage directions.
An interesting thing about Montessori is that every lesson is not only scripted but scripted with stage directions.
Sit to the right of student and use your right hand so as not to block the student’s view. Move the materials in this exact way. Say only this, don’t say that.
And yet…
…Montessori teachers almost never complain that this scripting is inadequate or creatively stultifying. They love it, use it, and depart from it cautiously and with reverence and with the guidance of hundreds of strong, fully explicated model lessons.
This is pretty legit. A lot of what was published recently about harness RL by @huggingface seems to track with what @GoogleResearch has done.
harness² from @GoogleResearch offers techniques for RSI on the harness, with a fixed model
they call it "harnessing", but you might already be familiar w/ this process! it's also called "hill climbing" or "continual learning"
more in 🧵 !
Patrick Skinner - edu/acc retweeted
Update - I'm getting distracted.
Unrelated: Marcy Stein is my co-author in Teach Your Child Foundational Math in 100 Easy Lessons.
This is the answer to getting your child engaging in math before school starts teaching them mis-rules through random exposures to toys.
Call me childish but today had me feeling like the 10yo kid in a comic book store that I once was.
My daughter came to me asking me to make her like Pepper Potts in Avengers: Endgame for Halloween. But, she wanted to glow in the dark.
I used AI to find some open source STL files for the Mark 49 Rescue Suit, resize them for her head and send them over to the 3D printer.
Meanwhile, I watched a car fly today.
I feel like we’re all witnessing, or maybe even living in, the arc of a Marvel comic.
Patrick Skinner - edu/acc retweeted
The year is 1897, horseless carriages are denied permission to be on roads by Washington D.C.
The year is 2026, people are trying to deny driverless carriages on the roads of Washington D.C.
Me hitting traffic on MoPac:
“Hey @grok, let’s make Tony Hawk proud.”
Patrick Skinner - edu/acc retweeted
we’re so back
Yo, Grok @bot, will I be able to voice chat with you on this thing?
Apple Homeview is official. $299
Going hands-on today, so ask me anything here!
apple.com/newsroom/2026/10/m…
Patrick Skinner - edu/acc retweeted
Some mistakes you might be making:
Your effort level is too high
Your skills are too long
You're not aligning before you start work
You're aligning too often, and too hard
Your automated checks are too lax
Your custom tooling is too token-inefficient
You aren't doing enough automated review
You aren't reviewing previous coding sessions (with an agent)
You're doing too much human review
You aren't vibe coding enough prototypes
Your agent can't verify its own work
You're babysitting the agent too much
You've overloaded your global steering files
Your codebase is hard to navigate
You aren't building a shared language
Your codebase architecture allows bad tests to be written
You aren't thinking strategically enough
You aren't thinking enough
I have all of this plus my agents can control my 3D printers.
Patrick Skinner - edu/acc retweeted
Even if subject knowledge was easily imparted across a system of teachers (it’s not) there is only a tenuous connection between knowing your subject and being able to decompose, sequence, and assess it effectively. This is why the field needs to embrace well-designed materials.
Patrick Skinner - edu/acc retweeted
A few days ago, I posted that “EdTech has had a cataclysmic shift, but companies haven’t noticed it yet.”
This video is a great overview of that shift and why software companies need to rethink their business strategy as soon as possible.
In education, I envision systems that aggregate the coming flood of options and build an individualized learning path for each student.
