I work for a US bank and I have observed that my job has shifted more towards creating Agentic workflow (fancy name of using LLM to automate tasks). In the last one year, I haven't touched any ML model. I am curious to know what is the experience of other folks.
I work for a US bank. I am in the Data Science team and for the last year I haven't built any ML model. Most of my work requires me to analyse unstructured data using LLM, automate using LLM API, and a little bit of simple LangGraph based workflows.
Curious to know the experience of other people working in the industry.
I work for a US bank and I have observed that my job has shifted more towards creating Agentic workflow (fancy name of using LLM to automate tasks). In the last one year, I haven't touched any ML model. I am curious to know what is the experience of other folks.
I've got an old TV hooked up to my Mac over HDMI for movies and shows. Every time I wanted to pause, skip, or turn the volume down, I had to get up and walk over to the keyboard. I downloaded Remote Mouse to fix this. The trackpad/mouse part is free, but media controls and volume were locked behind a ₹79/month subscription or a ₹1,499 lifetime unlock. So i built my own and is sharing with others.
It's a small menu bar app for macOS. You click the icon, it starts a local web server, and you open the Mac's IP address in your phone's (any phone with a browser) browser, nothing to install on the phone at all, just a web page. From there:
Security, since this is a remote control for an actual computer: pairing needs a 6-digit code that only ever appears on the Mac's own screen, plus an on-screen QR code if you'd rather not type the IP by hand. The code expires in 60 seconds, and only one device can be connected at a time, anything else trying to connect gets blocked until you disconnect the current one from the menu bar. No accounts, no cloud, nothing leaves your LAN.
For anyone curious how it's built: a Go server handles the web UI and logic, Cgo-linked directly into a compiled Swift object file for the actual macOS calls CoreAudio for volume, dlopen/dlsym into Apple's private DisplayServices and MediaRemote frameworks for brightness and now-playing info (there's still no public brightness API on Apple Silicon), CGEvent for simulating mouse and keyboard input. The whole app is under 9MB, no Electron, no bundled browser.
Full honesty, since people will ask: I have a data science background, not formal software engineering training, I'm learning as I go, and this project has been part of that. The Go server, the architecture, and the pairing/security design are mine, written first by hand and cleaned up afterward with an LLM's help. The Swift side, Apple's undocumented frameworks, the Cgo bridge is basically all AI-written. Swift isn't a language I know well enough to have done that cold, and getting AI to bridge it got me there a lot faster than working through disassembled framework headers myself would have. I'm not going to pretend otherwise. If you spot something dumb in either half, I'd genuinely like to hear it that's most of why I'm posting this instead of just keeping it on my own Mac.
Free, MIT-licensed, on GitHub: https://github.com/adarsh9780/mac\_remote.
Install it via Homebrew or build from source both are in the README
The blood you see was just after 2 hours after the surgery when they sent me home. I didn’t touch it all, no bump no accidents.
The rest of the photos are of today. Surgery happened yesterday on 11 june, by 7 i was out. Reached home by 8.30.
It is almost 24 hours now.
Did i fuck something up?