Setting up a weekly study rhythm: 2 hrs a day, one subject each day.
Math, quantum physics, computer engineering, economics, and psychology on weekdays. Sanskrit and poetry on weekends.
Long game, not a sprint.
Day 1 of economics (MIT 14.01). Core idea so far: scarcity forces trade-offs, and every choice has an opportunity cost. Also learned the difference between positive analysis (how things are) and normative analysis (how things should be).
Day 1 of linear algebra (Strang's 18.06, MIT OCW). Row picture, column picture, elimination โ three different ways to solve the same system of equations.
@flavioAd lmao, i am making something similar but for apple watch because i want to see biological age and AI features on my data but dont want to pay for bevel, UI is in shambles for now but gets the job donw
Next up: ESP32 sensors for AC control + water tank level via Home Assistant. Then a custom voice that isn't generic Siri.
Full writeup with the architecture, tools, and what's next:
https://t.co/0uZPhkTCrH
"Hey Siri, what's my CPU temp?"
"Fifty seven degrees."
Spent the weekend giving my homeserver a voice.
Apple Watch โ FastAPI โ @GroqInc โ 5 tools โ spoken reply.
No app. No dev account. ~250 lines of Python.
The 1050 Ti in the server never touches the reasoning loop.
Groq is faster and free; the GPU keeps doing NVENC for Jellyfin and goes toward Frigate later.
Use cloud LLMs for the brain, keep your local GPU for what it's actually good at.