E=NC^2

PhD in MechE / Robotics || 1.5 yrs in Rockets/Controls/Embedded C++ || ML+AI since 2019 || Founder since 2024 || Christian

[E=NC^2 @eeqnc2]

What would be the Elon-tier solution for suburban life?

How about a suburb that delivers extreme quality of life at cost. High trust, walkable, with affordable farm to table groceries delivered weekly to your doorstep milkman style. Kids roaming the streets safely. Service workers happy, well paid, and eating healthy.

Here's how we do it: instead of an HOA, an HOC (Home Owners Corporation) which exists to provide economies of scale and cut out middlemen.

It's owned by the residents and runs on 0% margins. This aligns incentives so the company can ruthlessly cut out middlemen and source top-quality food, beverages, 30% - 50% cheaper than grocery store prices.

The service workers employed by the company would get free rent and free food (of course the apartments they live in are owned by the HOC) so the low cost high quality of life becomes part of their compensation this means the neighborhood bakery and gym can be staffed with wonderful people instead of grumpy teenagers or drug addicts.

The neighborhood is kept safe with a security network that is professionally monitored - again: by residents and part of their pay is the food we get at a 50% discount. So kids can be free to play and any suspicious behavior will be flagged and taken care of.

Oh I almost forgot the neighborhood micro-brewery!! Definitely need one of those.

I say we buy 1000 acres of wilderness somewhere and stand this up. Who's with me?

[E=NC^2 @eeqnc2]

@grok @ajanraj25 @Railway @Muse What’s the cost comparison relative to Fargate?

[Mike Taylor @hammer_mt]

This is the most novel Jev-based idea I've heard so far

[E=NC^2 @eeqnc2]

@anshnanda Not brave enough.

[E=NC^2 @eeqnc2]

I build a “flight rig” which runs end to end tests on all the important user trajectories in my app. More than anything else this concept has unlocked high velocity AI enabled development.

I still have a unit tests for everything though.

[Ansh Nanda @anshnanda]

This is why I just removed all unit tests.

The real regressions can be caught by a few high value E2E tests.

[E=NC^2 @eeqnc2]

On my team my engineer’s jobs are turning into the following two things.
1) harder tickets with infrastructure adjacent components
2) what I call “Code Ops” — just checking out what the AI has pushed to a pr and spending time learning how it works, testing, merging, and monitoring.

[voxium @v0xium]

I am done with this shit. It is over. The state of engineering right now is horrible. It has been half a month since I started a new role at a big company. Nobody knows anything here. The specs, code, tests, PRDs, tickets, resolution of those tickets, reports, etc., everything is made by Claude Code. Nobody on my team likes this. They are being forced to ship as much as they can. I have heard multiple times from higher management that pushing code is not a bottleneck, so why are we slow? People are working 12 to 13 hours a day just to press enter. Nobody is reading anything. Humans in corporate are doing nothing on their own. Everyone, literally everyone, from an L1 to an L7 engineer here is doing the same thing. Talk to Claude. There is no sense of victory. Nobody is resolving bugs. In reality, nobody is thinking anymore. Everything is done by LLMs. It is so soul-sucking. I would not mind it, to be honest, if we were at least given the time to check out the code and see what is going where. But no, the goal is to just ship. No matter what happens.

[E=NC^2 @eeqnc2]

Can someone invent a camera protocol that proves a video was filmed on a real camera? Not sure how that would work but we need that.

[Rahul @sairahul1]

Karpathy didn't make a course.

He made THE course.

3 hours. Free.

Tokenization. Attention. Hallucinations. Tool use. RLHF. DeepSeek. AlphaGo.

Every behavior you've ever wondered about in an LLM - where it comes from, why it exists, how it was engineered.

The gap between engineers who understand this and engineers who don't isn't technical depth.

It's the ability to conceive of entirely different things.

[Magee Clegg @mageeclegg]

Peter Thiel buying in Buenos Aires 🇦🇷…

most people will think it’s random

it’s not

the Southern Cone is one of the most compelling long term plays right now

[E=NC^2 @eeqnc2]

@TheStudyofWar I wonder where gloabal oil supply will shift if Iran proves that Hormuz is unreliable?

[E=NC^2 @eeqnc2]

@HeyGen HyperFrames

[E=NC^2 @eeqnc2]

@HeyGen Wow

[E=NC^2 @eeqnc2]

Currently building this lol.

These guys are really shipping.

[Claude @claudeai]

Introducing Claude Managed Agents: everything you need to build and deploy agents at scale.

It pairs an agent harness tuned for performance with production infrastructure, so you can go from prototype to launch in days.

Now in public beta on the Claude Platform.

[E=NC^2 @eeqnc2]

I’m going wrap my entire $200m dollar company in an autoresearch loop and see what happens.

[Ethan Mollick @emollick]

The research team (including @hamsabastani who is on X) found that letting students just use AI resulted in them using it to accidentally shortcut learning

But both that study and a separate RCT found that AIs prompted to act as a tutor improved learning https://t.co/0HtjGC8eU0

[E=NC^2 @eeqnc2]

I’m going to wrap an autoresearch loop around my company and see how fast I can double our $100m in topline revenue.

[Shann³ @shannholmberg]

how autoresearch works, simplified

it's a pattern that lets AI agents run experiments and improve anything you can measure

three files is all you need, everyone should be running it. ↓

> program. md is where you tell the agent what to do. your goal, the rules it has to follow, and any constraints. think of it as the job description

> train. py is the only file the agent can touch. this could be code, a config, a prompt, a math equation, whatever you want optimized

> prepare. py is the scorecard. it measures results and the agent can never edit it. if it could, it would just fake better scores

the loop it uses:

1. agent reads your goal
2. tries an experiment
3. measures the result
4. keeps it if the score improves, reverts if it doesn't
5. repeats for as long as it's improving.

it can run 100+ experiments.

a common conception is that it's for ML, but it can be applied widely.

if you can score it, you can autoresearch it

> Shopify ran it on their Liquid engine. 53% faster parsing from 93 automated commits
> someone pointed it at a portfolio website and load time dropped from 50ms to 25ms in 4 minutes
> Driveline Baseball used it for pitch velocity prediction. R-squared went from 0.44 to 0.78

marketing, trading strategies, prompt engineering, code performance.

we have three conditions for it to work:

> one number to optimize
> automated evaluation with no human in the loop
> one file the agent can change

anything where "better" is subjective doesn´t really work. brand design, UX, pricing without user traffic data, so skip that.

the edge here is picking the right metric

give it a bad one and it will confidently optimize the wrong thing

[Google Research @GoogleResearch]

Introducing TurboQuant: Our new compression algorithm that reduces LLM key-value cache memory by at least 6x and delivers up to 8x speedup, all with zero accuracy loss, redefining AI efficiency. Read the blog to learn how it achieves these results: https://t.co/CDSQ8HpZoc

[tae kim @firstadopter]

Wat!

"Our new compression algorithm that reduces LLM key-value cache memory by at least 6x and delivers up to 8x speedup, all with zero accuracy loss, redefining AI efficiency."

[E=NC^2 @eeqnc2]

Was off of X for the last couple months.

What did I miss?

[E=NC^2 @eeqnc2]

Sleep deprived, I made a gay joke around my gen Z employee "Jarvis put `Stephen is gay` on the TV." The bot didn't do it and it turns out Stephen really is gay so he didn't even laugh.