ms. bubbles retweeted
i have a folder on my computer named "keep going"
here are ALL of the images inside of it: jackfriks.com/keep-going
ms. bubbles retweeted
Built a small Markdown renderer utility tool for windows to render markdown on terminal through PowerShell including Latex for maths
code: github.com/aadityansha06/win…
ms. bubbles retweeted
A small video to Understand Computer architecture especially Low-Level with in-depth intuition Behind CPU, Cache, memory hierarchy and locality and refresh overhead.
The video went quite long otherwise would have covered more advance topics
youtu.be/A1MiPITlYlc?si=Q78q…
ms. bubbles retweeted
Replying to @magic_bubblez
tabs stacked up until twenty-six sessions were open and only three were doing any work. the other ones were idle until i got back to them, which is why i built nuphos
github.com/nuphos/nuphos
ms. bubbles retweeted
launching kiban
a storage engine i wrote from scratch in rust. it read 100M records in 57s. RocksDB took 7 min
cached point read: 1.09 µs. 2.6x RocksDB, 1.3x LMDB 16 writers, every write synced: 14,101/s. RocksDB 12,830
it wasn't always like this. kiban once returned 8 when the answer was 40. sync() was reporting success on data a power cut would've wiped, and snapshots were losing data. my bugs. they're all written up on the site
so i tested it harder. simulated power loss, millions of crash seeds, kill -9 and reopen 1000 times
files never change after they're written so each one gets a perfect hash. you compute where the key is instead of searching for it. LMDB can't do that, its pages change in place
RocksDB still beats it on long write runs by about 3x, and on memory and disk. linux only, no transactions, not production ready. warm cache, my laptop
every benchmark has a receipt and a button that copies a prompt so your llm can explain it. tell me if it finds something
kiban-website.vercel.app
p.s. see the site on laptop, i didn't optimize it for the phone, it'd be a major disservice
ms. bubbles retweeted
i built weft, a cloud coding agent, to learn what it takes to make one trustworthy.
point it at a github repo and ask a question or request a change. it searches the code with grep and glob, edits in an isolated docker sandbox, reruns your tests in a clean one, and opens the PR.
ms. bubbles retweeted
We now have DSA sheet on Trentorch.com
- No you don't have to pay for anything
- 450+ questions
- 20+ Topics
- Multiple Languages: C++, Py, Java, JS, Go, Rust
- Theory of all topics
Learn and practice across Arrays, Strings, Linked Lists, Trees, Graphs, Dynamic Programming, Greedy, Tries, and much more.
-> We don't charge our users anything.
-> It's completely Opensource
-> No Paywall
ms. bubbles retweeted
I just asked Claude to reverse engineer Claude's entire growth strategy.
It called @treg_ai and used their brand new REA growth skill to find me literally everything the brand is doing in terms of growth right now.
Distribution is still the moat, buts its getting super easy to copy.
The future of GTM is going to be crazy.
🤝 Paid partnership
Introducing /REA-growth skill
Reverse-engineer any company's growth
- Name any company, get its growth blueprint
- Every move on its timeline: SEO, AEO, ads, social
- $0.58 for a full Lovable teardown
- Fully OpenSource
Try it at treg.to/rea-growth
Partnered with @sharenow_today
Repo link below 👇
ms. bubbles retweeted
okay so marin v0.1 is finally here 🥀
after way too many cuda kernels, memory bugs and endless optimization loops, i can finally show what i've been building lol
marin is my own local llm inference runtime written in rust + cuda. no llama.cpp or python underneath.
the whole idea was to make local inference usable on regular consumer hardware without needing some crazy setup.
it has cpu/cuda inference, int8, streaming chat, context reuse, batching, persistent sessions and even a tiny background assistant that can suggest help based on approved activity.
here's a lil demo of me actually using marin from the terminal.
some benchmarks and repo link below ↓
ms. bubbles retweeted
We're living in an economy where a guy with a master's degree is delivering groceries to a 22-year-old who made $4 million filming himself screaming at video games. And we're still telling kids to study hard.
ms. bubbles retweeted
This paper answers:
why anything is beautiful
why jokes are funny
why we get curious
why you get bored
why babies explore
why music has hooks
For centuries, we thought these were deeply mystical human traits. Soul-level mysteries that machines could never touch.
But, the paper proves that the human brain operates as a relentless compression engine. It is constantly trying to predict the world and compress it into the smallest possible file size.
Here is how it explains your entire life:
If something is completely predictable, your brain has already compressed it. There is no new data to extract. The result? You feel bored.
If something is pure randomness, your brain cannot compress it at all. The result? You feel annoyed or tune it out.
But then there is the sweet spot.
When you encounter something that seems complex, but suddenly your brain finds the hidden pattern and connects the dots... your brain releases dopamine.
That sudden, glorious rush of successful data compression? That is what you experience as "curiosity."
When a comedian tells a story that makes no sense, and the punchline suddenly makes all the pieces fit together instantly? That rapid neural compression is what you experience as "humor."
When you look at a painting and notice an underlying symmetry that ties the chaos together? That is "beauty."
Art, science, music, and jokes are not magic.
They are just highly optimized data sets designed to trigger compression progress in your neural network.
Here is the most unsettling part.
This paper wasn't written by a philosopher. It was written by an AI pioneer to figure out how to program intrinsic motivation into artificial agents.
He wasn't trying to explain humans. He was writing the blueprint for machines.
We always assumed AI could never appreciate art, laugh at a joke, or feel the spark of curiosity.
But if our deepest subjective experiences are just a mathematical reward for maximizing compression progress...
ms. bubbles retweeted
A single subscription of Clay or Apollo can cost you more than $100!
I built an artifact in Claude that runs @treg_ai with Jev to find the right buyers for my products for less than a dollar.
It takes me less than 10 seconds to use people search to find the right buyers, and Jev classified and scored them for less than $0.0004.
Super cool launch! Excited for what they cook next!
🤝 Paid partnership
Introducing Jev @typesafeai powered people-search
Just $0.0089/lead, No more $600 subscriptions,
Try it at treg.to/people-search
- 85% cheaper than Clay
- #1 on people search bench accuracy
- Plugin to any agent to find anyone
Fully open source, 0% markup
Git Repo below 👇
ms. bubbles retweeted
Today we introduce Heckler by @MercityAI
Heckler stress tests your AI agents with simulated users and shows you exactly where your agent breaks the rules you gave it.
We ran 3 agents (DeepSeek V4.1 Flash, GLM 5.3 and GPT-6 Luna) on Heckler and over 80% of chats broke a policy.
ms. bubbles retweeted
Today, we officially launched our AI Governance offering across our socials, and we’ve already been running a campaign around it for our clients.
Couldn’t have picked a better time, lwk, with half the internet seemingly getting compromised lol.
If your enterprise has been feeling the heat lately, AGH might just be the way to go.
Whether you’re worried about AI leaking sensitive company data, employees using unapproved AI tools, or AI budgets getting out of hand, this one’s for you 😼
ms. bubbles retweeted
Lots of Security incidents today, here's some practical tips for those who just wanna protect yourself:
1) UPDATE EVERYTHING - main targets are:
a) Browsers - Chrome, Safari, Firefox
b) OS: : Settings -> General -> software update; Linux: sudo apt update && sudo apt upgrade; Windows: I think there's a button. Do this for any remote home servers & Openclaw/hermes boxes
c) Phones: Update it all
d) Desktop apps (Slack, Cursor, Discord, etc)
2) Limit web browsing this week: Stay within walled gardens (twitter, youtube, etc); don't click random links this week. Just wait until the patches go through. Do step 1 again after a week.
3) Avoid untrusted Wifi networks until your devices are updated
4) Stay vigilant about phishing emails offering "certificate fixes" or fraud updates
5) Have an iPhone 11-16? TURN BLUETOOTH OFF for the time being until apple patches come through.
6) Code/Dev: UPDATE ALL YOUR REPOS - @charliermarsh has already updated uv 0.13.4 which reacts to the eMudra incident for uv (Python), Typesript according to Claude has equivelant. Tell your agents to go and update all your active repos to update package manager across uv, npm, bun, cargo, etc etc
What advice did I miss?
ms. bubbles retweeted
ok, i finally got around to thinking hard about what the biggest drawback of AI coding agents like Claude and Codex is...
it’s not them making mistakes or making developers dumber.
imo, it’s how they’ve turned this into an unfair fight.
AI has narrowed the visible gap between a newcomer, a junior engineer and a senior engineer by a ridiculous amount, while somehow raising the level required to get in.
as a newcomer, you could always jump in, work your way through code and literally outwork the other person.
but now the baseline has been raised so much that your productivity also depends on how many tokens and subscriptions you can afford. like, the capability bottleneck that decides where you'd end up is still 70% of the game, with tokens being the other 30%, but in a cutthroat industry where even 1% matters, that's a lot.
non-tech people can build shit insanely fast now. so just being able to build something is no longer enough to call yourself a software engineer. the bar keeps moving higher.
this is suspiciously becoming pay-to-win.
it feels like you're not even allowed to take your time and work through HTML, CSS and JS by hand anymore because the FOMO is so strong. while you're learning how everything works, people using better agents are already shipping.
for someone passionate who likes to romanticise the little things, this is a huge drawback.
maybe the weird thing is that this is in the process of becoming a layer we don't see anymore, like bits and memory allocation eventually became.
getting your foot in the door is becoming harder.
the path from newcomer to junior engineer is somehow harder but also much simpler now: keep grinding DSA and keep building cool shit until someone lets you in.
i don't think the actual gap between a junior and senior engineer is disappearing. experience and judgement still matter way too much.
but the material keeps moving down.
distributed systems and system design were always labelled as things you'd study alongside your job after 4–5 years, while trying to move towards senior.
now me and a lot of other college students are already working through them because the bar is getting so damn high.
what's frustrating about this is that you're kinda expected to choose your domain before you've worked in it in real life.
you chose backend? ok, now become competent. sure, you grinded, and now you're really competent. now you'll get a job.
wdym you don't like it? too bad, you can't switch because you'll have to go through that process all over again.
this is one place where i believe DSA and big tech are still sorted. you can still get into a good job using a tool that will help you build your fundamentals on the job and choose later.
FAFO, which imo is the core of this field, is becoming tougher at the career level. you can build a lot more things, so you can FAFO, but what happens when you realise this isn't the playground you wanna play in?
it has never been easier to build software, but getting your foot in the door somehow keeps getting harder.
especially when the amount of money you can spend on AI directly affects how fast you can move.
and the biggest thing: once we've fully embraced handing our execution over to AI, we pay these companies somewhat reasonable money rn, but what happens when they pull the rug?
why is my future in an industry also dependent on what a billion-dollar corporation decides to price its product at tomorrow?
i get that the industry is in a transition phase, but this whole thing feels weird...
there are a lot of firsts in this industry that we've never seen before, and that's what's kinda worrying me.
just my lil two thoughts and what I've been reading into.

























