marv :|

Dr. Pharm. in view || conceptual. || pivoting -- technical writer. || stablechain?

[marv :| @honmarveled]

somewhere between the noise of ai hype and gpu flexes, a new problem quietly emerged.

how do you use these massive models without letting them use you?

large language models (llms) have evolved insanely fast.

what started as simple text generators are now powerful reasoning machines.

capable of writing code, summarizing research, building agents, and even managing workflows.

but there’s a catch. for the longest time, the best llms were completely centralized.

owned, trained, and controlled by a few massive players.

so if you wanted access, you had to go through them.
you had to send them your data, your prompts, your context, everything.

in short, you paid with both your money and your privacy.

then things started shifting.

open-weight models began surfacing.

first llama, then qwen, and lately the monster, deepseek r1.

for the first time, people could actually run these models themselves, fine-tune them, build their own systems on top.

the gates opened.

but here’s the problem most people don’t talk about.

these models are huge. like, truly huge.

deepseek-r1, for instance, packs 671 billion parameters.

just holding those weights in full precision would need about 1.3 terabytes of gpu memory.

to actually use the model for prompts? easily double that.

now imagine this:
an nvidia h100 (the gold standard right now) gives you 80gb of vram and costs around $25k.

to run inference for a 2.6tb model, you’d need about 32 of them, that’s over $800,000 just on gpus.
and that’s not even counting networking, power, or infrastructure.

so yeah, “running your own llm” sounds great on paper.

in reality, most people (and even most orgs) can’t.

the result? third-party inference services.

they host these open models for you and run the prompts on their hardware.

you get access, they get your data.
and just like that, we’re right back where we started.

the same privacy trade-offs, just dressed in open-source clothes.

so the real question becomes:
how do you make llm inference private, even when someone else is running it for you?

that’s where this series comes in.

over the next few posts, we’ll unpack how privacy-preserving llm inference actually works, why it’s so hard, and what new methods are being tested to solve it.

and we’ll start with one of the oldest, yet most promising approaches.

secure multi-party computation (smpc).

be there tomorrow...........

Credit: @ArkaPal999

[marv :| @honmarveled]

@Johnobinnabrig1 hmm, tell me more.

[marv :| @honmarveled]

hi

[marv :| @honmarveled]

lmaoo, i bought terbinafine topical from you recently.

[Modupe🌺 @OneModupe]

Get in jor🎉🎉🎉
After 5 years of learning under my boss, i now have my own pharmacy store…
Tell me congratulations 🍾

[marv :| @honmarveled]

@Najeeb_STV i don’t think people realize the impact of this problem. we’re going to face a major increase in the unemployment rate, not because the labour market is inefficient but because there is simply no need for them.

i think the only way we can tackle this problem is to up skill.

[NAJEEB 🦅 @Najeeb_STV]

Architects! 🤝

@samconnerone x Najeeb.

[NAJEEB 🦅 @Najeeb_STV]

gm ct

have you tested clario on @arc? @samconnerone and @bobbilee needs to see this.

app link: https://t.co/m99AEPzYK4

[marv :| @honmarveled]

@Najeeb_STV motion boy!

[NAJEEB 🦅 @Najeeb_STV]

https://t.co/c1wpTcKDex

[NAJEEB 🦅 @Najeeb_STV]

clario is coming to @arc soon.

i’ve got school work, but this should be ready for testing before the end of the week we’re entering tomorrow.

just a few ui/ux fixes left.

anticipate, frens.

[marv :| @honmarveled]

i no be mavo i no dey okay, owo mofe ko ma plenty.

[marv :| @honmarveled]

@sopeee_ee mehnn what 😭

[Japa AI @justjapa_ai]

Your migration roadmap in 3 minutes. Fill your background. Get matched. Unlock your report. https://t.co/iMHKIdmRMB

[marv :| @honmarveled]

@jalaal_tweets this is great jalaal.

[$JRD! @jalaal_tweets]

Last week I embarked on a research on DAO Treasuries and what I found is shocking;

$16.4 billion in DAO treasury assets sitting in volatile native tokens with zero hedging.

85% of DAOs hold their entire treasury in a single crypto asset. In the last bear market, that meant 70–90% drawdowns.

The only automated tool serving mid-tier DAOs? Shut down in 2025.

So I built Wardex, a solo hackathon submission to @SuperteamUK “Campus to colosseum”

It's an AI agent that reads on-chain treasury data, forecasts runway, scores risk, and benchmarks DAO treasuries against each other.

Built for the DAOs with $1M–$50M that Karpatkey and Llama will never serve.
Demo live → https://t.co/F9MocVhC8H

[Kakashi @kkashi_yt]

When you buy a memecoin coz no one is selling then realise it was because no one can sell.

[Blackhaven @blackhaven]

In just 5 days, 45,000+ users participated in our Early Access Campaign.

To filter out sybil farms and bots, we've partnered with @getmoni_io to ensure that only real participants get their points boost.

We'll be sharing more on what's next in the coming days.

[marv :| @honmarveled]

Help me qualify for (3,3) on @blackhaven to get early access. Sign up through my link and grab yours before it's too late.

https://t.co/ejeW5QuVU2

[Raven🤭 @RavenX09]

Since December 2025, interns at @OAuthc have not been paid salaries, with multiple arrears outstanding.
Many of us rely on this income for basic survival.
This is really unfair
4 months salary outstanding😭😭😭😭😭😭😭

[Elnathan John @elnathan_john]

https://t.co/Bv6hgN8Ttd