Kategorien durchsuchen
Entdecken
Fiverr Pro
Deutsch
$
USD
I will deploy private ollama chatbot on linux server
Daniel
Über diesen Service
Want a private AI chatbot without sending data to OpenAI or paying monthly fees?
I will deploy a private, high-performance Local LLM chatbot (Ollama) on your Linux serverbare-metal or containerized.
Hardware & OS Support:
Nvidia (CUDA), AMD (ROCm), Apple Silicon (Metal), CPU-only.
Ubuntu, Debian, Fedora, Arch, Alpine, RHEL.
Native Systemd OR Docker / Docker Compose containers.
What I Offer:
Basic Package:
Ollama install (Native or Docker).
1 Open-source model (Llama 3, Qwen 2.5, DeepSeek).
CLI & local API verification.
Standard Package (Recommended):
Everything in Basic.
Web UI Interface (Open-WebUI) for easy browser chatting.
24/7 Autostart (Systemd unit or Docker restart policy).
Custom system prompt configuration.
Premium Package (Production Ready):
Everything in Standard.
Nginx Reverse Proxy + free HTTPS/SSL setup.
Multi-model deployment & GPU passthrough.
Secured API auth tokens to connect external apps.
Requirements:
Linux server access via SSH.
For Premium (SSL): Valid domain pointing to your IP.
Message me before ordering!
Lerne Daniel kennen
Daniel
Python Automation and AI Systems Engineer
- AusDeutschland
- Mitglied seitAug. 2026
- ⌀ Antwortzeit1 Stunde
Sprachen
Deutsch, Englisch, Polnisch
Software engineer specializing in Python automation, AI deployment,
and Linux server infrastructure.
What I deliver:
• Private AI chatbots on your Linux server (Ollama, no cloud)
• Python scripts, bots, scrapers, and data pipelines
• REST APIs (Flask, FastAPI) and backend development
• Linux, Nginx, Docker, and CI/CD setup
• C# tools, plugins, and build pipelines
Clean, documented code with setup instructions.
Delivered on time. Message me before ordering!
Meine weiteren Dienstleistungen im Bereich KI-Entwicklung
FAQ
Do I need a GPU on my server to run Ollama?
High-core CPUs (AMD EPYC, Mac M-series) run 8B models fine (5-15 tokens/s). Low-core cheap VPS CPUs will be very slow (1-3 tokens/s). For instant ChatGPT-like speed, you must use a GPU.
Do you need root (sudo) access to my server?
No. I prefer user-space installs for security. I can install Ollama in ~/.local/bin via systemctl --user. Sudo is only required if you want Nginx on port 80/443 (Premium).
Is this actually 100% private?
Yes. Ollama runs entirely on your local metal. Zero telemetry, no cloud API calls, and absolutely no data is ever sent to OpenAI, Anthropic, or any third party.
Can I integrate this into my existing apps?
Yes. Ollama exposes an OpenAI-compatible REST API. Point your existing OpenAI apps to http://127.0.0.1:11434/v1 and use local models without changing any underlying code.
Does this work on Docker?
Yes. I deploy bare-metal (Systemd) or containerized (Docker). If you have Nvidia GPUs, I will configure the Nvidia Container Toolkit (--gpus all) for hardware passthrough.
What happens if my server restarts or reboots?
Ollama will automatically start up on boot without any manual intervention. I configure background autostart using Systemd service units or Docker --restart always policies.
Can you customize the AI's behavior or persona for my business?
Yes. In the Standard and Premium packages, I configure custom system prompts and parameters (temperature, context window) so the AI speaks in your company's tone or follows your exact coding standards.
How do auto-updates & container maintenance work?
For Docker setups, I can configure Watchtower to automatically update your Open-WebUI containers in the background, or pin specific version tags if you prefer strict stability and zero breaking changes.
How are service health & logs monitored?
I set up HTTP healthcheck endpoints (/api/tags) and configure systemd/Docker journal logging with automatic log rotation to ensure your server doesn't run out of disk space from log bloat.
How do custom system prompts work under the hood?
I write custom Ollama Modelfiles (FROM qwen2.5-coder, SYSTEM """...""") to lock in your exact business persona, system instructions, temperature, and context length directly into the model binary.
