







My ₹38k Intel N150 homelab setup - Debian, Jellyfin, Immich, Tailscale and portable 5G Wi-Fi
I recently (About a year) upgraded my home server to a Skullsaints Ash mini PC based on the Intel N150. I wanted something compact, low-power and easy to leave running.
It shipped with Windows 11 Pro, but I wiped it and installed Debian on bare metal. I had been running an Arch Linux machine as server earlier, so I reused the HDD from that machine for media storage.
Specs
- Intel N150
- 16 GB RAM
- 512 GB SSD
- Around 1 TB reused HDD for media
What I paid
| Item | Price | Bought From |
|---|---|---|
| Skullsaints Ash mini PC, Intel N150 / 16GB / 512 GB | ₹24,999 (Oct 2025) | Amazon |
| TP-Link TL-WR1502X | ₹3,728 (June 2026) | Flipkart |
| Qualcomm SDX55 5G USB dongle | ₹8,258 including tax (June 2026) | The Techtellectual |
| Ambrane Stylo N20 20,000 mAh power bank | ₹1,275 (July 2026) | Flipkart |
| Kitchen rack | ₹186 (July 2026) | Flipkart |
Total: ₹38,446, excluding the reused HDD and SATA to USB 3.2 Connector.
I replaced my D-Link DIR-600M with the TP-Link TL-WR1502X partly because the old router had become a bottleneck for Wi-Fi and wired speeds. But the main reason was that the TP-Link can run from a basic power bank.
Paired with the SDX55 5G dongle and the Ambrane power bank, it becomes a portable Wi-Fi setup that I can take outside. It gives my phone and other devices their own Wi-Fi network without requiring a wall socket. I can also keep my phone’s cellular radio off when I want to while still using internet through the router and 5G dongle.
Docker Stack
- Jellyfin
- Immich, PostgreSQL, Redis and Immich ML
- Radarr, Sonarr, Lidarr, Prowlarr, FlareSolverr, qBittorrent Web UI and Seerr
- Tailscale for remote access
- Caddy and DuckDNS — currently stopped while I rework the configuration
- OpenRouter proxy
- Hermes Agent and the Hermes web dashboard
- SearXNG and Camofox Browser for Hermes
I only use legitimate media sources.*
iPhone apps I use with it
- Hermes Conduit
- Swiftin iOS
- Immich
As I am adding AI agent Stack with cloud models, Figuring out the best way to use the N150 without running out of RAM. Suggestions are welcome, especially from anyone else using 5G as part of their homelab network.