Hands-on setup help
Self-Hosted LLM Setup Service. Fixed price, fixed scope.
Choose the right hardware, get a working stack, and leave with a checklist you can run yourself.
Quick Call
$79USD / one time
A 30-minute call before you buy hardware or commit to a stack.
- A written hardware and model shortlist
- Monthly cost assumptions and trade-offs
- A clear next step: DIY, setup or wait
No server access or deployment.
Target: within 48 hours of a confirmed booking.
Start with a callStandard Setup
Core package$299USD / one time
One working stack on one prepared machine.
- One agreed runtime, Web UI and model
- Versioned configuration and evidence
- A 60-minute walkthrough and handoff
No multi-GPU, clusters or integrations.
Target: 3 business days after prerequisites are ready.
Discuss your setupProduction Setup
$749USD / one time
Single-host setup with a defined hardening and test scope.
- Reverse proxy, HTTPS and API access control
- Monitoring, load report and one tuning pass
- Deployment, rollback and handoff notes
No Kubernetes, fine-tuning or uptime SLA.
Target: 5 business days after prerequisites are ready.
Review your requirementsService fees only. Hardware, GPU rental, API usage, domains and applicable taxes are separate. Availability, prerequisites and final scope are confirmed before payment.
Three costs. One service fee.
Free software does not remove infrastructure costs or the time needed to configure and maintain it.
- Software
- Use the agreed tools and model under their own licenses. Paid software or model licenses are not included in the setup fee.
- Infrastructure
- You own and pay for the machine, GPU, storage and network. Any rental or paid API usage needs your approval and is billed to your account.
- Setup and support
- The one-time fee pays for the agreed work and handoff. It does not include ongoing hosting or open-ended operations.
What you receive
Each delivery is checked against an agreed list. The handoff should let you repeat the setup without depending on a hidden configuration.
| Deliverable | Quick Call | Standard Setup | Production Setup |
|---|---|---|---|
| Written recommendation | Shortlist, budget and next step | Agreed stack and prerequisites | Agreed stack and operating assumptions |
| Deployment record | Not included | OS, hardware, drivers and exact versions | Same, plus access and network diagram |
| Repeatable configuration | Not included | Compose or runtime config + secret-free env template | Same, plus proxy, monitoring and rollback config |
| Test evidence | Planning assumptions only | Successful request, screenshot and single-request baseline | Baseline + agreed load test and one tuning comparison |
| Walkthrough | 30-minute planning call | 60-minute handoff; recording by consent | 60-minute handoff including operations notes |
| Limited follow-up | No deployment support | Two same-environment follow-ups within 30 days | Two same-environment follow-ups within 30 days |
The runtime, model, quantization and acceptance conditions are agreed before payment. Recording requires your consent; a written walkthrough can be used instead. A performance report is a measurement of your setup, not a model-quality guarantee.
Where the scope ends
The service delivers an agreed technical setup and a usable handoff. These boundaries keep the quote predictable.
Hardware sales, GPU purchasing and server hosting are not included. You keep ownership of the infrastructure and its ongoing bills.
Do not send API keys, passwords or private keys through this form. Enter your own secrets. Any necessary access uses an agreed temporary account that you revoke at handoff.
No fine-tuning, data labeling, custom business-system integration, multi-node clusters or Kubernetes work is included.
Model accuracy, hallucination rates and business outcomes are not guaranteed. Acceptance covers the agreed technical checks on the agreed environment.
Work beyond the package is quoted at $60/hour and starts only after written approval. No automatic extra charges.
Follow-up covers problems in the delivered environment. A different model, operating system, hardware or integration needs a new scope. No 24/7 monitoring or uptime commitment is included.
Confirm fit, then pay
You do not need to pay to find out whether your machine and requirements fit a package.
- 1
Describe the machine and the blocker
Send your hardware and VRAM, operating system, target model or task, and what is currently failing. No credentials or private logs.
- 2
Agree on the package and the checks
Review hardware readiness, runtime compatibility, availability, delivery date and acceptance criteria. If the machine is not ready, start with a call or prepare it first.
- 3
Pay through Waffo after scope confirmation
Receive the agreed payment request. The service fee is paid before work begins. No checkout charge is created by this page.
- 4
Run it yourself at handoff
Receive the configuration and evidence, then repeat the agreed workflow using the document. Record any remaining issues and the follow-up window.
Before you request a setup
A few practical questions about readiness, access and what happens after delivery.
Can I buy a setup before I have the machine?
Start with the $79 Quick Call. Choose the hardware and confirm compatibility before paying for deployment. A setup slot starts only when the prerequisites are agreed and ready.
Do you support macOS and Apple Silicon?
For this service, the Mac scope is a compatible native Ollama or llama.cpp path. Linux GPU setups are assessed separately. Do not assume that a Docker container on a Mac can use Apple Metal acceleration.
Will you need my passwords or API keys?
No long-lived secrets should be sent by email or through this form. You enter your own keys. If access is needed, agree a temporary least-privilege account and revoke it at handoff.
What if the agreed setup does not work?
The quote must state the technical acceptance checks and cancellation/refund terms before payment. If the agreed setup cannot be delivered, pause the work and resolve the issue or refund under those agreed terms. This page does not replace applicable customer rights.
Can my company purchase the service?
Include the billing requirements in your inquiry. Available receipt or invoice details and taxes are confirmed with the payment request; do not assume that a specific local tax invoice is available.
Can we communicate in Chinese?
Yes. You can send your requirements in Chinese or English. Confirm the language of the call and handoff document when agreeing the scope.
Tell me what you want to run
Five short answers and a package preference are enough for a fit check. Generate a brief, review it, then open an email draft.
Do not paste passwords, API keys, tokens, private IPs or unredacted logs. “Not purchased yet” is a valid hardware answer.
Contact: janus.choy@gmail.com