Junel retweetledi
Junel
138 posts
Junel
@junthedev
Primarily DevOps. Sometimes backend, but can dabble into blockchain too. I like anime, video games, and pipelines.
Katılım Kasım 2024
82 Takip Edilen32 Takipçiler
Junel retweetledi
Junel retweetledi
9 years ago, i watched a guy get fired for doing the right thing... he built the team, trained everyone, fixed the messes, stayed late, took no credit. the kind of employee companies claim they value.
then he questioned a policy that was quietly screwing junior staff, didn’t insult anyone, didn’t shout. he just asked why... two weeks later: HR invited him and served him some papers saying he’s not a culture fit... then laptop seized, access cut. goodbye.
the manager who created the policy got promoted, the company posted a linkedin thread about psychological safety... bro lost his job for speaking up.
in most companies and workplaces, ethics are tolerated until they challenge authority... you can be loyal, competent and ethical but none of that matters once you threaten someone’s power, ego or comfort.
the system will protect itself first, always and people are replaceable but control is not.
elyshia@elyshianone
Hit me with the harshest reality truth.
English
Junel retweetledi
Junel retweetledi
You can now run 70B LLMs on a 4GB GPU.
AirLLM just made massive models usable on low-memory hardware.
𝗪𝗵𝗮𝘁 𝗷𝘂𝘀𝘁 𝗵𝗮𝗽𝗽𝗲𝗻𝗲𝗱
AirLLM released memory-optimized inference for large language models.
It runs 70B models on 4GB VRAM.
It can even run 405B Llama 3.1 on 8GB VRAM.
𝗛𝗼𝘄 𝗶𝘁 𝘄𝗼𝗿𝗸𝘀
AirLLM loads models one layer at a time.
Instead of loading everything:
→ Load a layer
→ Run computation
→ Free memory
→ Load the next layer
This keeps GPU memory usage extremely low.
𝗞𝗲𝘆 𝗱𝗲𝘁𝗮𝗶𝗹𝘀
• No quantization required by default
• Optional 4-bit or 8-bit weight compression
• Same API as Hugging Face Transformers
• Supports CPU and GPU inference
• Works on Linux and macOS Apple Silicon
𝗪𝗵𝗮𝘁 𝘆𝗼𝘂 𝗰𝗮𝗻 𝗱𝗼
• Run Llama, Qwen, Mistral, Mixtral locally
• Test large models without cloud GPUs
• Prototype agents on cheap hardware
English
Junel retweetledi
Someone curated 925 failed VC-backed startups, broke down why they failed, and how to make it work with today’s tech -
loot-drop.vercel.app
Cool fr🙌
English
Junel retweetledi
One summer evening 20 years ago, my dad came home from work with a new PS2 game, it was Ace Combat Zero. I loved it so much that my dad then bought me 4 and 5.
Ace combat is the connection to my childhood and my late dad.
Thank you for the early Christmas gift @BandaiNamcoUS.
Bandai Namco US@BandaiNamcoUS
Fall of Wings. "Wings of Theve" has fallen, but the legend lives through you. From now on, you’ll carry the title and be the greatest ace pilot this world will ever know. ACE COMBAT 8: WINGS OF THEVE is coming to PS5, Xbox, and PC in 2026. Wishlist now: spr.ly/6019CDV5J
English
Junel retweetledi
I literally just studied the Nginx Ingress Controller a week ago 🥹
Akhilesh Mishra@livingdevops
Kubernetes is retiring the NGINX Ingress Controller by March 2026. And the internet is having another meltdown. "NGINX is dead!" "Kubernetes is abandoning us!" It's the Docker panic all over again. NGINX, the web server, is completely fine. It powers half the internet and isn't going anywhere. It's the Ingress NGINX controller that is retiring. This is one specific Kubernetes project that uses NGINX under the hood to route external traffic into your cluster. Why is it happening? The Ingress NGINX project has been running on fumes with just one or two people maintaining it in their spare time. Security holes kept showing up. They tried building a replacement and begged for help. Nobody came. So instead of pretending everything’s fine while your clusters are at risk, Kubernetes is being honest and saying, “This thing can’t continue safely, please move on by March 2026.” Unfortunately, even that announcement failed to generate additional interest in helping maintain the project. What should you do? Move to Gateway API. It's more powerful, better maintained, and where the ecosystem is headed. Cloud providers are already adopting it by default. Or switch to another Ingress controller like Traefik or HAProxy. Many support both Ingress and Gateway API, giving you time to migrate properly. You have options. You have time. Don't panic. it This is progress, not the end of the world. Read the full breakdown here: @akhilesh-mishra/nginx-ingress-controller-is-retiring-and-no-nginx-isnt-dying-b67a166b36de" target="_blank" rel="nofollow noopener">medium.com/@akhilesh-mish…
English
Junel retweetledi
Junel retweetledi
Junel retweetledi
Junel retweetledi
MOJANG IS REMOVING OBFUSCATION FROM JAVA!!! PEAK
English
Junel retweetledi
A curated list of free courses with certifications.
English
I tried using Outlook, I quit in about 3 months. I still think Gmail is superior.
alex fazio@alxfazio
the state of microsoft in 2025, building the same app three times in c#, c++, and typescript
English
Junel retweetledi
This is the first time we've ever posted pornography.
English
Junel retweetledi
You should try Elysia
> 1 line OpenAPI
> 1 line OpenTelemetry
> 1 line cookie signing + secret rotation
> 1 line End-to-end type safety
> 1 line Bun Fullstack Dev Server
> Built-in validation + Standard Schema support
> Best in class Type Safety + Soundness
> Anime girl mascot
English
Junel retweetledi
