i've been running qwen 3.6 35B A3B with llama.cpp on a 3090ti. i have found it better then sonnet in many ways. Speed and iterations was key. here is the gist of my current configuration: https://gist.github.com/spencerkittleson/5e44b6895a17ca45161... I use this with tailscale so all my devices have full access to it. That machine get toasty....
I made a bunch of silly projects at first to help me manage daily life. For example, automating saving important emails based on prior history. Extracting text and putting it into csv files, downloading finance reports, reminder scripts, watchdog scripts. All the boring stuff pretty much what was done at first.
https://thespaceabove.us/ extremely good listen with a lot of details in the early years of the space program. I found renewed interest in the area. Even started replicating some of the missions in KSP
https://docodethatmatters.com It was a way for me to build, optimize, and enrich my skills around the web standards. Also, a place for me to document some of the fun projects around DSLR camera hacking, Raspberry PIs, home automations and 3d printing
One thing I really like is the performance... its smooth and fluid. The api is really useful as well: I wrote a small job to auto add descriptions and tags to the images.
Moved from nodered only to a hybrid of nodered and home assistant. Added some new sensors, nfc tags, modes and automations for multiple tenants / cost savings. Its been fun to automate some boring tasks.