Devember 2026: Home Lab Edition

I remembered this thing was going on so why not sign up as well, I’m making a small music player that plays local tunes for me in a way I like and will run on anything I manage to get my hands on. It has tooling so I can attach it to an AI agent and manage playing from there.

The learning part is actually not about the program at all and the program is more of a byproduct if anything. It’s about everything that surrounds the actual program, the pipelines, version control, the tooling running multiple platform builds, the corner cases, when to optimize, what to optimize, can we force feed it meatloaf and how will it react, and generally how far can we take it before things start to break down.

3 Likes

This is my idea: Running Compiler Explorer with a local project , don’t know if it’s even a good idea :grimacing: . Let’s see how far I get. I’ll have a public repo soon.

3 Likes

That is a very cool idea

1 Like

Oh thank you! This is the first time I’ve shared the idea beyond 1 friend. I have tried out local aspects of the idea with curl as a client, seems to be quite okay :slightly_smiling_face: , now to the disassembly part :crossed_fingers:

(I’m not sure if my project qualifies; it’s a software project, heavily built using a clanker, and parts already exist, in fact this project started as a devember project in 2022)

For this Devember 2026 I am reviving my Server Control Panel in the look of Windows XP: SCPxp, and extending it beyond reason, using a Clanker, into a networking/VM playground running inside the browser.

It was once a simple front-end for common shell tools in the browser, with a simple CGI backend to run any command(e.g. the file manager uses tree, cp, etc., the terminal emulator is tmux).

With the help of a Clanker I’ve re-written it to make it more functional and pretty:

  • It now hosts multiple VMs in the browser, using v86.
  • Complex networking setups, with a complete network stack implemented in the browser(Layers 1-7: audio modem, virtual switches, full TCP/IP using lwIP, an HTTPS server and a Firefox port for WASM).
  • networking can integrate with external “real” networks via tailscale(DERP), websocket relays(host yourself) or WISP servers(public ones available)
  • It can also connect to other SCPxp networking instances via WebRTC(peerJS for signaling).


Screenshot is showing: a network plan with an in-browser OpenWRT VM, reachable from an outside browser via tailscale, another VM connected to the OpenWRT switch, and a bunch more network devices in the plan connected to another network.

This is my second Clanker-coded hobby project(my first clanker-enabled project is a toolchain for my Minecraft CPU including an LLVM port). It’s meant as a learning exercise, not a serious production project!
I just wanted to see what is possible using the Clankers, and learn something about network etc. while doing it.

Getting this released and onto my website/GitHub in a useable form is the actual devember project.
The rewrite is, as of now, completely unpublished.
It’s still missing a lot of polish(a lot of the UI text definitely reads as LLM-generated), a Linux image that ships the backend CGI server and panel(so everyone can try the full SCPxp without requiring an installation, just by visiting a website), some apps should be separated out into separate repos(e.g. the Minecraft CPU related stuff), - That’s what I’ll be documenting for devember.

I’ve been posting some progress in this thread as well(even before Devember 2026 started): https://forum.level1techs.com/t/what-are-you-working-on-today/

Not sure if my project even qualifies, but some human feedback would be nice regardless!

Can’t wait to see more of the other devember projects!

3 Likes

I want to set up set up an AI workspace (mostly not for coding) that can leverage a few of the GPUs I have at home as well as cloud models but chat history and files stay local while I can chose what I run locally and what gets to the cloud models. with RAG, voice, etc. Inspired by Techno Tim’s I Built My Own Intelligent AI Gateway video.

PCIe lane adapter galore and making a hybrid AI workspace

4 Likes

I like this!

My November project unfortunately doesn’t quit align with homelabbing, so I’m posting my September project. The goal was to boot the FritzBox and plan the networking for the new house. I settled on used Sophos gear mainly for the pricetag. The router of choice could be used for more than just a router. Linux fully supports its hardware so you can boot whatever.

While I didnt assign any specific budged, I try to stay below 100€ for any equipment that I ordered.

This setup is running for almost two months now :slight_smile:

Tightening my tin-foil hat. To quote Wendell “I don’t care how paranoid you are you are not paranoid enough. It’s not paranoia if they are out to get you!!”

I’m the Moolokki. I like linux :penguin: I have thought about this for some time now, (possible) remote access to my devices home w/o DDNS (writing this, I’d guess tor/i2p works too…) & routing my internet traffic from my phone/laptop to originate from my home. That’s cool, I am of the opinion that tech serves you, not the other way around.

I am aware that what I have achieved could be done using tailscale or netbird.

I have a second project that I’m working on that’s related(ish), I wish to setup frigate using an old IP-Camera that’s flashed with OpenIPC. Next steps for that are backup the OG firmware and flash it.

I’m so intimidated by how smart most people’s projects are for this month. Out of my friend group, I think I am consider one of the tech, smarties… and I see so much here I want to learn from.

Anyway I originally thought about doing something like Turnstone and trying to network my pcs that are so old… and I may still add that as scope creep that if I have time, but I went with something that I think should be simple and easy:

Putting NFC tags around my house to help me log when I do maintenance or trigger reminders through home assistant.

LLM stuff feels like it could be a forever tinkering project, but I feel like I could learn and deploy some real nfc solutions in the next month. Start and finish.

I’m afraid that I’ve forced myself to overhaul my devember ideas, because I’m replacing my NAS :sweat_smile:

I’ve been looking at upgrading/replacing my current NAS for a while… But with the Beelink ME Pro kickstarter which @wendell made a preview about, I decided that the N150 304 (I changed my mind :sweat_smile: ) 4-bay model perfectly suits my storage needs.

Guess I’ll have to expand my project list with a TrueNAS build and a couple of services I have running on the old model. (Which will probably be placed off-site as a back-up at my mum’s house.)

My Devember project is Clustara, an iOS/Android app for managing my homelab from my phone. SSH, Proxmox, Docker, terminals and logs, all connecting directly over LAN or VPN.

I’m a sr devops engineer who’s spent most of my career on web apps. I started thinking about this in late October and finally had something I wanted to build for mobile. I don’t write my own code anymore; Claude Code wrote almost all of this.

I’ve also been using a crazy Frankenstein GPU cluster for local AI: an RTX 5090, two RTX PRO 4000 Blackwells, two RTX 5080s and an RTX 3090. That’s 136GB of VRAM across two machines, connected with 100GbE/RoCEv2. It’s running DeepSeek V4.1 Flash at Q2_K_XL through llama.cpp RPC with a 256K context window. A fairly unreasonable amount of hardware to end up talking to a terminal, but I’ve been getting plenty of use out of it. but the TTFT is brutal but after warm cache its not bad

My goal for Devember is to get it working on more labs than just mine and uncover all the it works on my machine bugs

1 Like

A better Remote Access Protocol

I see lots of people responding with already finished projects. This one is mostly just an idea I’ve been thinking about the last week or two.

I’m thinking of calling it something like: “A better Remote Access Protocol“. Basically a “better ssh“ (although as a stretch goal also screenshare).

(post deleted by author)

I’m being a bit ambitious with goals for this, but most of these are projects I’ve been planning/gathering materials for already.

I plan on building network security/privacy nodes, introducing some of my family to Meshcore off grid messaging, building an emulator PC that will eventually be using an iMac G3 for the CRT, and hopefully refreshing my media PC to get off Windows/Plex

Below is what I am shooting for, though my stepfather passed away last week, so I am out of town with my mother for who knows how long. I have full SSH and WebUI access to the machine I will be testing on, but physical access is relegated to how often my son is down to help be my hands and eyes. I built this machine back in Q12026 and spent several months tooling a custom PowerShell assistant from several million .md files from the usual PS places, of note, MS directly. It has worked well and helped me automate a lot of things in my daily work flow for the better without worrying about token costs. But the main reason is anyone that has used the big players to work on intricately detailed PS projects they just are not that good. Even CoPilot I would say is the worst. But I jays2cents. With all of the INSANE advancements in tooling for the ROCm and R9700 platform specifically of late, I figure why not build what I really need. Outlined below is the local AI assisted on my current build goals for this Devember. Again, if I fall short with everything going on with the family life thing, apologies for wasting your time. Admittedly the goals are lofty to say the least, but not entirely impossible.Without further ado:

Darth Omniscience

A Local-First AI Homelab Operator on Dual Radeon AI PRO R9700

Mission

My Devember 2026 project is to learn how to turn a dual-R9700 Linux workstation into a local-first AI operator for my homelab.

The goal is not simply to run a large language model. The goal is to build a system that can observe the state of my infrastructure, correlate telemetry across otherwise separate systems, explain failures in useful technical language, and eventually perform tightly controlled actions on my behalf.

Examples of the questions it should ultimately answer include:

  • Why is this NAS not responding?

  • What is the uptime and current load of that Docker server?

  • Does that Windows host appear to need a reboot?

  • Which containers are unhealthy?

  • Why has a requested movie not reached Plex yet?

  • When was a monitored license plate last observed by UniFi Protect?

  • Which smart-home devices are unreachable?

  • What changed immediately before a service failure?

Project Philosophy

This is a learning project first.

AI may assist with research, code review, debugging, documentation, and boilerplate, but important architectural decisions, configuration, testing, troubleshooting, and integration work must remain understandable to me.

I should be able to explain:

  • why each component exists;

  • how data moves through the system;

  • what permissions each service has;

  • what happens when a dependency fails;

  • how a decision or action can be audited;

  • and how to recover the system without AI assistance.

Core Design Principles

Local first.
Inference, telemetry, automation, logs, and household data should remain local wherever technically practical.

Observe freely; act narrowly.
Darth may receive broad read-only visibility into the homelab, but write operations will be explicitly allowlisted.

No unrestricted root shell for the model.
The language model will not receive direct root SSH, unrestricted Docker socket access, full Home Assistant administrator access, or blanket UniFi administrator credentials.

Tool broker between AI and infrastructure.
Infrastructure actions will be exposed as discrete tools such as:

get_host_uptime(host)
get_container_health(host, container)
pause_download(client, id)
restart_allowlisted_container(host, container)
set_home_assistant_entity(entity, state)

The broker, not the model, defines what is possible.

Read → Suggest → Approve → Automate.
New capabilities begin read-only. They progress through Shadow Mode and human approval before any policy-driven autonomous action is considered.

Everything important is logged.
Tool invocation, arguments, requesting user, result, errors, approval state, and execution time should be auditable.

Monitoring must survive Darth being offline.
Persistent telemetry and logs should live on lower-power always-on infrastructure rather than depending exclusively on the AI workstation.

Recovery before cleverness.
No change to kernels, GPU drivers, storage, networking, or remote access is accepted until rollback is understood.

Primary Hardware

Darth Omniscience:

  • AMD Ryzen 9 9950X3D

  • 128 GB system RAM

  • 2 × AMD Radeon AI PRO R9700

  • 64 GB aggregate VRAM

  • Ubuntu Linux

  • Dedicated AI/model/data storage

The first technical phase will benchmark the current R9700 software ecosystem rather than assuming a single inference runtime is best.

Candidate runtimes include:

Radiance
Paiton
vLLM/RDNA4 variants
Strata
llama.cpp
AMD-supported ROCm inference stacks

The architecture will be chosen from measured results on this hardware.

Planned Functional Areas

Infrastructure observability

Linux hosts
Windows endpoints
ZimaOS systems
TrueNAS/Synology storage
Docker containers
UniFi networking
UPS/power state
system logs and metrics

Home automation

Home Assistant
approximately 90 WiZ lights and sensors
local-only device control wherever possible
inventory, location, naming, firmware state, and health

IT management

NinjaOne/NinjaRMM read-only telemetry initially
pending reboot and patch state
device health and alerts
future allowlisted remediation

Media automation

Plex
Sonarr
Radarr
Overseerr/Jellyseerr
Transmission
related storage paths

Security and cameras

UniFi Protect
camera health
event history
ALPR event research
local retention/querying of useful event metadata where supported

Remote Access

Private administration will not require direct public exposure of the homelab.

A small client application, provisionally called DarthConnect, will eventually:

  1. determine whether Darth is securely reachable;

  2. establish the approved VPN path if required;

  3. validate the service endpoint;

  4. open the Darth interface;

  5. report connection state clearly.

UniFi WireGuard, Teleport, Tailscale, or another appropriate private-access method will be evaluated.

Devember Judge Access

Judges will not receive broad VPN or LAN access.

If a live demonstration endpoint is provided, it will use a separate Judge Mode with sanitized telemetry and restricted read-only tools.

Private systems such as cameras, ALPR data, household automation, NAS contents, NinjaOne endpoints, and unrestricted infrastructure controls will remain inaccessible.

Default Assistant Personality

The default delivery profile is Captain:

calm
measured
precise
authoritative
dryly understated when appropriate
British in cadence

The intent is a distinguished Shakespearean-command style rather than a literal imitation or clone of any real actor.

Shadow Mode

Before Darth receives meaningful write authority, new integrations operate in Shadow Mode.

Example:

User:
“Why is Zima2 behaving badly?”

Darth:

Diagnosis: HandBrake has been continuously CPU-bound for 47 minutes and its worker stopped producing output 19 minutes ago.

Recommended action:

restart_container("zima2", "handbrake")

Shadow Mode:

No action performed.

Only after repeated correct recommendations will a tool graduate to approval-required execution.

Success Criteria

A successful Devember project does not require every planned integration to be complete.

The core project succeeds when Darth can reliably:

  • observe several independent homelab systems;

  • correlate their telemetry;

  • answer useful operational questions;

  • explain how it reached its conclusions;

  • recommend a remediation;

  • execute a small set of explicitly authorized actions;

  • record what it did;

  • remain remotely recoverable;

  • and demonstrate meaningful use of the dual R9700 hardware.

The project should leave behind not only a working system, but documentation sufficient for someone else to understand the architecture, reproduce the important pieces, and learn from the failures encountered along the way.