<div>
 <strong>I want to make powerful local AI easier for everyone to use.</strong>
</div>
<div>
 <br>
</div>
<div>
 I'm raising money for an NVIDIA RTX 5090—or a comparable high-end GPU—to help me continue an open-source, educational project around local AI.
</div>
<div>
 <br>
</div>
<div>
 This started as a personal, off-hours project.
</div>
<div>
 <br>
</div>
<div>
 I wanted to understand what it actually takes to run modern AI models yourself: on a Mac, on Windows, on inexpensive hardware, and eventually on increasingly capable local machines.
</div>
<div>
 <br>
</div>
<div>
 The more I learned, the more I realized that the technology is becoming incredibly powerful, but getting started can still be unnecessarily complicated.
</div>
<div>
 <br>
</div>
<div>
 So I'm trying to make it simpler.
</div>
<div>
 <br>
</div>
<div>
 <strong>What I've been building</strong>
</div>
<div>
 <br>
</div>
<div>
 I've been writing practical articles and publishing open-source repositories that explain how to run LLM servers locally across different platforms.
</div>
<div>
 <br>
</div>
<div>
 Some of the projects I've already built include:
</div>
<div>
 <br>
</div>
<ul>
 <li>Mac / Apple Silicon: an MLX-based LLM server</li>
 <li>Windows / NVIDIA: a local LLM server setup</li>
 <li>Raspberry Pi: experiments with running LLM infrastructure on small, inexpensive hardware</li>
</ul>
<div>
 <br>
</div>
<div>
 The idea behind all of them is the same:
</div>
<div>
 <br>
</div>
<div>
 <strong>Take something complicated, figure it out, simplify it, and share what I learn.</strong>
</div>
<div>
 <br>
</div>
<div>
 I don't have all the answers. I'm learning as I go, and that's intentional.
</div>
<div>
 <br>
</div>
<div>
 AI is changing extremely quickly. I think one of the best ways to understand where it's going is to actually build things with it, run the models yourself, measure what works, and share the results.
</div>
<div>
 <br>
</div>
<div>
 <strong>Why I need a 5090</strong>
</div>
<div>
 <br>
</div>
<div>
 I'd like to take the project much further.
</div>
<div>
 <br>
</div>
<div>
 A high-end NVIDIA GPU would give me enough local compute to experiment with substantially larger and more capable models and to explore workloads that simply aren't practical on the hardware I currently have access to.
</div>
<div>
 <br>
</div>
<div>
 I want to explore the full local AI stack:
</div>
<div>
 <br>
</div>
<div>
 <strong>text → vision → audio → image generation → video generation → agents → APIs → self-hosted applications</strong>
</div>
<div>
 <br>
</div>
<div>
 And then turn what I learn into articles, examples, and open-source repositories that other people can use.
</div>
<div>
 <br>
</div>
<div>
 There is something powerful about being able to download a project, run it on your own machine, and understand exactly what is happening.
</div>
<div>
 <br>
</div>
<div>
 I want to help make that experience normal.
</div>
<div>
 <br>
</div>
<div>
 <strong>Why crowdfunding?</strong>
</div>
<div>
 <br>
</div>
<div>
 The honest answer is that I can't afford a GPU like the RTX 5090 myself.
</div>
<div>
 <br>
</div>
<div>
 This is an off-hours project. I also have a 2.5-year-old son, so buying several thousand euros of computer hardware for a personal project isn't something I can reasonably prioritize.
</div>
<div>
 <br>
</div>
<div>
 At the same time, I don't want that to be the reason the project stops progressing.
</div>
<div>
 <br>
</div>
<div>
 So I'm asking whether the people who find this work interesting—or useful—would like to help fund the hardware.
</div>
<div>
 <br>
</div>
<div>
 This isn't a company fundraising round.
</div>
<div>
 <br>
</div>
<div>
 There isn't a business behind this.
</div>
<div>
 <br>
</div>
<div>
 It's one person spending his spare time learning, building, writing, and sharing.
</div>
<div>
 <br>
</div>
<div>
 And I'm hoping a community of people who are excited about local AI might want to help me keep going.
</div>
<div>
 <br>
</div>
<div>
 <strong>What happens if we reach the goal?</strong>
</div>
<div>
 <br>
</div>
<div>
 I'll buy the best suitable GPU within the campaign budget, ideally an RTX 5090 or comparable hardware.
</div>
<div>
 <br>
</div>
<div>
 Then I'll put it to work.
</div>
<div>
 <br>
</div>
<div>
 I'll continue publishing what I learn and building projects around local AI, with an emphasis on making them understandable and practical rather than overly complicated.
</div>
<div>
 <br>
</div>
<div>
 I'll also share the journey.
</div>
<div>
 <br>
</div>
<div>
 What works.
</div>
<div>
 <br>
</div>
<div>
 What doesn't.
</div>
<div>
 <br>
</div>
<div>
 How fast different models run.
</div>
<div>
 <br>
</div>
<div>
 What hardware actually matters.
</div>
<div>
 <br>
</div>
<div>
 How to deploy these systems.
</div>
<div>
 <br>
</div>
<div>
 And, most importantly, what ordinary people can realistically do with AI running on hardware they control themselves.
</div>
<div>
 <br>
</div>
<div>
 <strong>You don't have to donate.</strong>
</div>
<div>
 <br>
</div>
<div>
 If you like the idea, that's enough.
</div>
<div>
 <br>
</div>
<div>
 If you've learned something from one of my articles or repositories, a donation is one way to help me keep doing this.
</div>
<div>
 <br>
</div>
<div>
 A small contribution is useful.
</div>
<div>
 <br>
</div>
<div>
 A large contribution is obviously very useful.
</div>
<div>
 <br>
</div>
<div>
 A share is useful too.
</div>
<div>
 <br>
</div>
<div>
 And if you happen to have a suitable GPU that you're no longer using, I'd be very happy to consider a hardware donation as well.
</div>
<div>
 <br>
</div>
<div>
 <strong>Why I'm doing this</strong>
</div>
<div>
 <br>
</div>
<div>
 I think local AI is going to become increasingly interesting.
</div>
<div>
 <br>
</div>
<div>
 Not because everyone needs to own the fastest GPU.
</div>
<div>
 <br>
</div>
<div>
 But because people should have the option to understand, experiment with, and run powerful AI systems themselves.
</div>
<div>
 <br>
</div>
<div>
 The hardware will keep getting better.
</div>
<div>
 <br>
</div>
<div>
 The models will keep getting better.
</div>
<div>
 <br>
</div>
<div>
 The tooling will keep getting better.
</div>
<div>
 <br>
</div>
<div>
 I want to help make the path from "<strong>I've heard about this</strong>" to "<strong>I'm running it myself</strong>" as short and approachable as possible.
</div>
<div>
 <br>
</div>
<div>
 That's what I'm working toward.
</div>
<div>
 <br>
</div>
<div>
 If you'd like to help me get the hardware to take the next step, thank you.
</div>
<div>
 <br>
</div>
<div>
 I'll do my best to turn that support into something useful that others can learn from and build on.
</div>
<div>
 <br>
</div>
<div>
 And I'll keep sharing what I learn along the way.
</div>
<div>
 <br>
</div>
<div>
 GitHub: <a href="https://github.com/kibotu" target="_blank" class="" rel="nofollow noopener noreferrer">https://github.com/kibotu</a>
</div>
<div>
 <br>
</div>
<div>
 Article: Two Paths to Local LLM Servers — Windows/NVIDIA vs. Mac/Apple Silicon
</div>
<div>
 <br>
</div>
<div>
 Thank you for being part of the journey.
</div>

Anonymous

Help Me Build a Serious Local AI Workstation

platform

Platform is iOS

Android

Adyen Google Pay Development Audience

$opt_dummy_attribute

English only

Deprecate Stripe Legacy

dyn_template_ids

Adyen V6 Bene Transfer & MyMoney Campaign Organizers - test01

Bene transfer V6 and AdyenV6MyMoney Campaign Organizers - production

OneTrust next-app audience

fundCountryCode

NextDoor Share Option Users

Fundraisers with Reactions feature capability

fundCurrencyCode

US + Non-charity COs Only

locale

fund country is US and user language is English

personId

comms-internal-person-ids

Enhanced Story Experience Users

Suggested Goal Experience Users

user language is English and fund country: US, AU, DK, FI, IE, NO, SE, CA, GB; 

Poster share - new funds eligible (newest 9/18)

Streaming widgets beta user

paymentProcessor

3DS payment method

user language is English and fundCountry: US, AU, DK, FI, IE, NO, SE, CA, GB; 

GoFundMe Employees

ipAddress

Mark Rieth IP Address

Version 9.2.0 and above

Fund is in English speaking country

desktop

Android v12.11.0 and above

Platform is Android

paymentMethodType

checkoutType

Payment method type group to be vaulted on success

admin pr kit prod

EUR Donation Upsell

Share Text generation Skip Fundraiser

(prod) new funds eligible for amp12 (UGC gallery)

fintrac_cad_idv_form_v2 audience

paymentProcessorAccountId

fallback mode accounts

proCampaignId

p2p legacy startfundraising page

be_supporter_comments_enabled_supported_funds

AMP24 New Funds Only PROD (DO NOT MODIFY)

countryCode

charityId

UKCharities

Federated Search Internal Testing

route.pathname community

US Donation Upsell audience 

Auto Thanks Feature Audience

Supporter Comments Eligible Funds

International Only

fundId

braze fundraiser group 

Payment Identity Transfers ATO Risk Check (Targeted) - Production

Fundraiser Saved Payment Method Rate Limit Bypass (Production)

leaderboard-partner-id-audience-givepanel

Version 9.2.1 and above

AUD Donation Upsell audience

fundraiserCountryCode is US/EN

Active Fundraiser - Smart Default Cadence

DAF Internal

Vault on success card payment method

Platform is iOS v11.0.0 and above

Desktop users

Person ids

United States (US) users

allow yuno fundraisers for test

Mexico country code

Posters INTL

NPO Profile Opted-In

iOS at or above 12.4.0

non-US recurring donors

Canada Campaign Owner Group

Fundraiser Without Meta Link

Scheduler payment tester gorup

charityCommunityId

Charities allow to be claimed unclaimed (PRD)

Canada (CA) users

currencyCode

UC Tipping v20250115 International

pymk_production_xplr_fe_team

Android at or above 12.2.0

Profile photo entry point audience

smart goals (fundraiser/manage) on test01

Activity Tracking v1 Beta Testing

writePath

Stream Notification

start gfm intl allowed locales

Platform is iOS v11.2.0 and above

prod_ach_audience

AI Chatbot Audience

amp13 only new funds (prod) (anyword auto)

UC Flow Enabled (Prod)

Payment Identity Beneficiary Verification - Production

shared notification bell audience

Custom users

ach frontend audience

Android v11.0.0 and above

Users Receiving Payments Paused Notifications

smart goals test01

Manage Coach - US English Organizers Only + new funds only

Team EGG - IP Address list

Fundraiser Without Instagram Deep Link

Native 11.4.0 and above

UC Tipping v20250107

Users Receiving Braze KYC Validation Failed

Native 11.3.0 and above

Influencer - Prod

US fundraisers and US users

ai_title_page

anyword

openai

control

manual_entry

userEmail

template_id

userId

fundUrl

isCharityCo

email

appVersion

shouldBucketIntoExperiment

partnerId

donationCadence

isActiveFundraiser

isPartner

isMobile

npoProfileOptIn

fundIsLinkedWithMeta

hasSessionDonation

isCampaignOwner

fundHasInstagramDeepLink

Native 11.4.0 and higher

New funds eligible - Video Sharing (PROD)

GB Donation Upsell audience 

Auto Thanks V3 Rollout Audience

Thank you V3 testing whitelist

isRecurringFriendly

Is Recurring Friendly

Excluded funds - Video Sharing

Logged In Users

CH Donation Upsell audience

Android v10.6.0 and above

userCurrencyCode

Localized Currency

v11.5.0 and above

testAudience

pathname

Survicate

appVersion is at or above 12.8.0

Non-charity fund

isChallengeFundraiser

challenge-fundraisers

Payment Identity Launch Verification - Production

New funds eligible - Posters v2

CAD Donation Upsell Audience

Braze Email Tester

NPO Profile Opted-Out

iOS App Version

hasKnownDonors

PYMK has donors

iOS v12.3.0 and above

Help Me Build a Serious Local AI Workstation

Still €5,040 to go. Help us build momentum.