Kyle @ KyTech

Building in public 🚀 Roadmap/Feedback:

[Kyle @ KyTech @KyTechInc]

@kapish_dima I use the same stack on my VPS's, only edit would be Dokploy > Coolifyio

On my VPS's I block all inbound traffic via the vendor FW and use CF Tunnels to expose services.

[Kyle @ KyTech @KyTechInc]

@shadcn @sadmann17 🫶

[Kyle @ KyTech @KyTechInc]

@vanilagy Tested, results posted 🫡: https://t.co/CwOxlOdZL8

[Kyle @ KyTech @KyTechInc]

@larsencc nice! don't forget cargo clean for those tauri app build files that are huge!!!

[Kyle @ KyTech @KyTechInc]

@JordanAmblin @raycast That and even as a "Power User" leverage shortcuts or other automations its again tricky to target things without those icons. I often look at those icon as IOS app endpoints that can expose different functions.

[Kyle @ KyTech @KyTechInc]

@vanilagy Haha! ok good tip. For now i've just removed resolution scale and running only on a preset ladder.

This pipeline will be for a course/LMS platform so just looking for a uniformed way of standardizing user uploaded videos for quality, playback and tracking from S3.

[Kyle @ KyTech @KyTechInc]

@vanilagy Success! Only part im still tweaking is getting resizing working with some sort of canvas pollyfill

[Kyle @ KyTech @KyTechInc]

@vanilagy nice! ok giving this a shot!

[Kyle @ KyTech @KyTechInc]

@MaziyarPanahi @nvidia Right now I’ve played around with GLiNER a bit. Nothing in large sample size but in spot checks it seems to do well. I’m not native to think though there isn’t lots more room for improvement or stability in scale.

[Kyle @ KyTech @KyTechInc]

@MaziyarPanahi @nvidia Ya that would be awesome! I would be happy to test thoroughly. We have 40 years worth of data in the space that would be in scope for this type of processing or redaction flow.

[Kyle @ KyTech @KyTechInc]

@MaziyarPanahi @nvidia For our use we need to cover detections based on sensitive data under NIST and NERC CIP compliance standards.

Mainly this includes IP addresses, hostnames, subnets or other network / unique digital identity information that would be harmful to be leak even in isolation.

[Kyle @ KyTech @KyTechInc]

@MaziyarPanahi @nvidia Ok cool! I’m new to the training side of things but might give this a go!

[Kyle @ KyTech @KyTechInc]

@d4m1n For now.... lol

[Kyle @ KyTech @KyTechInc]

@leerob - Does @cursor_ai have some sort of reverse grandfathering in place for token usage/billing?

I have two accounts (1 personal, 1 at work) both are on Pro but on my personal which i've had much longer almost every request has 80%+ read cache hits. At work less then 10% hit the read?

I'm using Composer-2 on both accounts, working on the same type of codebases (new/old) and running the same Cursor versions (nightly stream on both)

Only difference i can see is on my personal in the desktop app in settings i have a "Plan & Usage" tab. But at work its missing?

Either way, I'm burning through tokens at work where it really matters and sipping on token usage for hobby stuff lol

[Kyle @ KyTech @KyTechInc]

@ctatedev GIS data next?