Frontend/fullstack (Vue, React, RN, Go) · building in public from NG
Nigeria
Joined October 2025
- Tweets1K
- Following359
- Followers142
- Likes2.7K
I think Machineward might be the slowest project I build, high-key.
My exams start this Saturday, and this is my first project that gets this close to how a computer actually works. I have a lot to learn as I build it.
I’ve been thinking about a problem I run into while building: I leave a long-running task on my computer, step away, and then wonder: is it still working, or does it need my attention?
looking into ideas for this my project literally, i don't really know what to design here, i'm blunt out of ideas, nothing is coming. appreciate anything from ya'll really.
This project is just building an automated "movie night" planner. It takes the media a user wants to watch and dynamically generates a menu of recipes that match the vibe, genre, or cultural origin of that specific movie or show.
I thought I had finally figured out how to control the internet speed on my PC.
I went into my MTN MiFi, found the Rate Limiting section and set my PC to:
Download: 500
Upload: 500
I ran a speed test.
1.66 Mbps download.
I was like, wait… what the hell? 😂
I literally told the router not to let my PC go past 500, so how am I getting 1 Mbps+?
I asked ChatGPT.
Turns out I was the one getting the units wrong.
The MiFi wasn't using Kbps.
It was using "KB/s"
So my "500 KB/s" limit is actually roughly "4 Mbps", not 500 Kbps.
That 1.66 Mbps I was complaining about was actually still below the limit I configured.
Nothing was broken.
I just compared "Mbps" from Speedtest with "KB/s" from my router without converting the units.
This is one of those small technical mistakes that look stupid after you understand them, but they're precisely the kind of things that remind me why understanding the fundamentals matters.
The numbers weren't lying.
I just wasn't reading them correctly.
This is the first version result guyy
taste-and-watch.vercel.app/
i don't how clueless i can be, i'm so clueless and confused rn
so basically this project is just building an automated "movie night" planner. It takes the media a user wants to watch and dynamically generates a menu of recipes that match the vibe, genre, or cultural origin of that specific movie or show.
It's a going to be single page for the MVP. but now i can't just build something and it will look like this when i'm done with the MVP, i'm trying to brainstorm the landing/search ui for the single page, nothing massive just to get the mvp on a smooth level, but nothing is literally coming to mind at all. need some crazy intervention rn.
yessss it workssss😭, the ai is back online and now it is actually generating the food pairings
so basically i was debugging this whole gemini integration thinking my api key was just straight up busted, but the first failure was actually a 404 from google saying the configured gemini-2.5-flash model is no longer available to new users, and explicitly instructing to use gemini-3.8-flash instead, so that actually was straight up understandable.
so i went ahead and switched the config over to 3.8-flash, which actually worked and the connection reached google instantly, but then it immediately returned a 503 high demand error(chai my brain wan blow😭). the second failure is materially different because the model endpoint was actually reached this time, but the provider was just temporarily overloaded, so the ai integration itself is fundamentally connected. this just means i need to stabilize the model and test path instead of randomly changing the architecture or thinking the code is completely broken.
i'm looking at how it was fixed by just bumping the model version, but getting slammed with a 503 right after is crazy. is there actually a way to skip the whole 503 entirely or just handle it better while testing? like another route to generally fix it the way i did but actually bypassing the overload completely?
