Ron Amosa
uncommonengineer.com
Ron Amosa
@uncommonengineer.com
💻 Hacker/Architect/Engineer.

🇼🇸 🇵🇸

Views are my own. Not speaking for my employer. Not professional advice.

Follow the White Rabbit ⬇️

https://www.uncommonengineer.com/
My tattoo artist brother always been the outspoken one, from local government corruption (see: ABA, Avondale), all the way to genocides on the other side of the world. Been saying it with his whole chest since the beginning ❤️ #FreePalestine 🇵🇸
August 17, 2025 at 8:31 AM
Sooo... the swamp on X is feeling some type of way today 🤣
July 27, 2025 at 5:24 AM
and you would be right... MedusaJS looks pretty siiick..

but also, dammit Ira! I'm already rabbit-holed on some other cool sh!t @santini.nz showed me yesterday 🤣
June 2, 2025 at 9:36 AM
I was happy to get my copy in the mail today, looking forward to reading and not being surprised by anything OpenAI has done, in this book 😂
May 26, 2025 at 9:50 AM
Reading RNZ, I'm often left thinking "wtaf is this piece saying or not saying and does it need to be this long to say it?"

controversial take: if this is human news writing, I'll take the AI summary, thanks.

www.rnz.co.nz/news/nationa...
May 14, 2025 at 6:58 AM
Jah. I know. I'm from this era 👇🏽 🤣

first thing I said to a dev who was explaining what MCP did for them, was ask them how this was different to these integration services from yesteryear that managed the back<->front communication flow.
May 11, 2025 at 12:22 AM
I have this weird thing where I get super curious when I see a new AI-related capability come out...

but I get legit excited when I see how that new capability can be pwned 😈

last week it was "look at this MCP thing I built"...

this week: hey, let's break MCP to pieces!

arxiv.org/pdf/2503.232...
May 10, 2025 at 2:55 AM
🤖 Finally had time to sit down and see what I could do with MCP, with one of the only use cases I find interesting with AI agents = hacker bots 😂.

Setup a local DVWA container, short python script worked, so onto a more sophisticated LLM agent next.
May 1, 2025 at 8:21 PM
👀

returning to this throwback for.... reasons.
April 30, 2025 at 8:41 AM
😂 why does this not surprise me 🇳🇿
April 18, 2025 at 10:54 AM
📘 Part 4 of 4 of my little deep dive into Agentic AI Security on LinkedIn.

Glad to get this one done. Deep dives are exhausting, especially once the initial excitement has faded and its just you, all your notes and your OCD to post it 😂😭.

Fin.

www.linkedin.com/pulse/part-3...
April 15, 2025 at 8:20 AM
🥷🏽 one for the hackers:

this was buzzy to see, in the open, on tiktoks website 🤔 ...

javascript RSA (jsrsasign) for API request signing (client-side), but for some reason, just letting the private keys ass hang out like this in public 😂.
April 10, 2025 at 8:30 AM
The results are in...
March 24, 2025 at 10:00 PM
I definitely spent waaaay too long on these infographics in Canva 😂😭 ... but they were fun to do.
March 24, 2025 at 8:59 PM
Finally got around to rebuilding my PiHole DNS w/ Unbound.

Cannibalized 3x HP thinclients to make the one supreme client, Debian 12, docker-compose, containers ftw.

Old school computer geekin'... twas a good day.
March 20, 2025 at 9:24 AM
Meta AI Studio, "Find and interact with AIs created by others." 👀

Other than the Quran being an obviously popular resource, the rest of the AI's (and this is just ONE section) doesn't bode too well for a certain er..."demographic" 😬
March 19, 2025 at 12:46 AM
So... basically PRE-PROD for human beings? 😂👀
March 17, 2025 at 9:58 PM
@anthropic.com’s latest model has landed… I like it.
February 24, 2025 at 8:38 PM
Of course the natural progression (in my line of work anyway) from containers is Kubernetes... And managed at that.

While containers are light, nimble, hackery and fun... Managed k8s is such a fucking slow death by comparison.

Enterprise, resilience, blah blah I get it... Still not fun.
February 13, 2025 at 9:49 PM
Man I love these little containers 😊
February 10, 2025 at 12:29 AM
This could be interesting 👀 - Multi-Agent AI System Prompt Infection Attacks 😈
February 8, 2025 at 9:38 PM
This piece by @404media.co hitting the nails right on the head 💯

https://s.mtrbio.com/yenfudnptj
February 7, 2025 at 12:10 AM
heh- this feels like a million years ago.
February 6, 2025 at 6:53 AM
what I did find funny was this: says DeepSeeks $6M training figure's either a lie or is not that cheap cos they trained Claude for 2-3x the cost?

"I won't give an exact figure"... so don't believe DeepSeek, but "trust you bro"?...
January 31, 2025 at 7:51 AM
Reading @anthropic.com's Dario Amodei's blog post in response to the #DeepSeek r1 model that has U.S. AI folks losing their shit.

At least he's up front about that 'American Imperialism' must prevail. For some reason he spells is "national security" later in the piece...
January 31, 2025 at 7:51 AM