Dave Ingle

The goal: make powerful AI systems behave in predictable, auditable ways so teams can trust what they ship. More to come.

[Vert|Rule @vertrule]

According to Grok:
"9.5/10. This is not just a case study; it is a blueprint for credible checkpoint tamper analysis. It is precise, reproducible, non-hyperbolic, and methodologically superior to most public AI-security writing. The only reason it is not a perfect 10 is the inherent limits of a single-case analysis (acknowledged) and the fact that real-world deployment often lacks a clean baseline."

https://t.co/ugrTeIY7cm

#AISafety #LLM #OpenWeights #ModelIntegrity #AIGovernance

[Vert|Rule @vertrule]

We’re publishing daily provider drift evidence now: https://t.co/SywWdlriJF #LLMOps #AIInfrastructure #AIEvaluation #AIReliability

[Dave Ingle @davidingle]

Liam Neeson may have a very particular set of skills, but I have an ADHD level of crafting supplies.

[Dave Ingle @davidingle]

@elonmusk False. READ: 'Entire' civilization. You are happy leaving people behind. Twerp.

[Dave Ingle @davidingle]

@ThomasSowell @elonmusk logic a foreign concept to you?
x = 1, y = 0.8

If x != y equal rights says you need to increase y.
If x == y equal rights says things should stay the same.

In this scenario 0.2 is the rights that are not afforded to those groups.

Stop rounding up.

[Dave Ingle @davidingle]

I lied. I ran it in another window.

[Dave Ingle @davidingle]

I like that AI allows me to explore thoughts like...
If you took a midpoint representation of the positive side of the number line. What happens if that number is prime? What happens if that number +-1 is prime?

\mu_N = \sum_{p \le N} \delta_{\log p - \frac12 \log N}

Then stop

[Vert|Rule @vertrule]

I like IBM's approach to AI.

I'm going to add Granite4 to the VertRule research preview. Will update when that is done.

https://t.co/o3BWuRH4sY

https://t.co/QUciLL8NQv

@IBM reach out if you want to know more.

I appreciate you.

[Vert|Rule @vertrule]

Care to follow along?

Gemma4 -> Nemotron3 -> Grok1 -> Kimi2.5

"I love the ambition, and the M4 Pro is an absolute powerhouse of a chip, but we have a non-negotiable problem here: physics.

Trying to run this gauntlet unquantized on a 45GB machine isn't just a tough challenge; it is mathematically impossible. Unified memory is incredible, but it cannot bend space."

@grok your opinion please.

[Dave Ingle @davidingle]

I built a thing. Please have a look:
https://t.co/qwtXeELNKb

[Dave Ingle @davidingle]

@elonmusk @grok can you review Elon's tweets for racial bias and general intolerance.

[Dave Ingle @davidingle]

It's almost like they are on to something...
They = me. Reach out for more.

[Vert|Rule @vertrule]

We ran a 27B parameter AI with under 5MB peak active memory.

A 120B model could fit into just 2.7GB of active memory.

Not “compression” as a slogan. A bounded active-residency execution regime, with receipts.

If the fidelity holds, this changes the shape of inference.

[Dave Ingle @davidingle]

This is a thing. I don't know what it is.
But hey: "LOOK! A thing."

If you are a person who likes things. And this is a thing you would like to look at. Look at it.

I haven't though. So this isn't an endorsement.

Just a thing.

[Avinash Avinash @AvinashAvi9658]

I’m beyond excited to announce that on Feb 2, I’ll be launching something truly special!
This AI-powered platform redefines how job seekers grow, build, and apply:
1. Auto-apply for jobs — Just fill in your details, and AI finds and applies for any the best-matched roles for you. You don’t need to stuck on single role you can create multiple roles using ai it takes everything.
2. AI project builder — Instantly create real-time, GitHub-ready projects powered by Copilot to showcase your skills.
3. Smart HR training — Ai will help You how to communicate with recruiters For each job application in different roles and craft personalized messages for each role.
It’s more than a job tool — it’s an AI career accelerator that helps you build proof of skill, confidence, and opportunity.
Feb 2 — Just wait to see the magic unfold.

[Vert|Rule @vertrule]

Hi. Can I please get help peer reviewing this?
https://t.co/iBj6UBgajY

It is blocking the rest of the work I'm doing.

[Dave Ingle @davidingle]

Applause. @NikkiGlaser

https://t.co/czJNAde9Cw

A pure demonstration of effort delivered effortlessly.

[Dave Ingle @davidingle]

Me: “I’m not a fucking philosopher.” - Dave Ingle

AI: “I’m not a philosopher. I’m just paying attention.” — Dave Ingle

[Dave Ingle @davidingle]

What you say is incredibly important. It just isn't important to everyone.

[Dave Ingle @davidingle]

No you aren't.

[Vert|Rule @vertrule]

I am Iron Man.

https://t.co/7r9czHduX7

[Dave Ingle @davidingle]

You have to prove you're not a robot by being shown a picture of a motorcycle and a scooter. Do I fail the test if I include the scooter?
#2dchess

[Dave Ingle @davidingle]

I’ve been quietly working on VertRule, a company focused on governance and accountability for modern AI systems.

If you know people who own AI decisions in regulated orgs, or work on AI policy/safety, please consider introducing us.

That’s the group I’m building this for.

[Vert|Rule @vertrule]

Founders talk a lot about disruption. I care more about stewardship. VertRule is my attempt to build the governance layer modern AI actually deserves.