frontpage.
newsnewestaskshowjobs

Open Source @Github

fp.

16-Year-Old Founded an AI Missile Startup. He's Targeting 'Cost per Kill'

https://www.wired.com/story/henrijs-dandzbergs-polemos-missiles-ai-warfare/
1•WithinReason•13s ago•0 comments

What's the environmental impact of my AI usage?

https://substack.hadley.nz/p/whats-the-environmental-impact-of
1•doener•56s ago•0 comments

Show HN: Is your favorite band in the Rock and Roll Hall of Fame?

https://hof.fyi
1•umilio•1m ago•0 comments

AI is imposing a hidden 'cognitive tax' on the lowest-paid workers

https://www.theguardian.com/technology/2026/nov/20/ai-low-paid-workers-cognitive-tax
1•Brajeshwar•2m ago•0 comments

El Niño Is Driving a Record Marine Heat Wave Off the California Coast

https://www.wired.com/story/el-nino-record-california-marine-heat-wave/
1•Brajeshwar•2m ago•0 comments

Show HN: Penumbra v2, an ergonomic colour scheme inspired by natural light

https://github.com/nealmckee/penumbra
1•solasluaith•2m ago•0 comments

Google's 3D emoji has wild Android spec requirements

https://9to5google.com/2026/11/19/android-3d-emoji-specs-requirements/
1•EbNar•3m ago•1 comments

Anthropic: After the IPO, Who Controls the Company?

https://www.morningstar.com/stocks/anthropic-after-ipo-who-controls-company
2•cybermango•3m ago•0 comments

New Programming Language Vine Based on Interaction Nets – InfoQ

https://www.infoq.com/news/2025/03/new-programming-language-vine/
1•rbanffy•6m ago•0 comments

Robert K. Brown, Publisher of Soldier of Fortune Magazine, Has Died

https://www.nytimes.com/2026/11/12/obituaries/robert-k-brown-dead.html
1•bookofjoe•6m ago•1 comments

Convenience Is the Actual Product

https://pawelbrodzinski.substack.com/p/convenience-is-the-actual-product
1•flail•6m ago•0 comments

Roman Numerals and the Symptom of Bad Design

https://pratik.deoghare.com/roman-numbers-and-bad-design.html
1•pratikdeoghare•8m ago•0 comments

America and Public Disorder

https://walkingtheworld.substack.com/p/america-and-public-disorder
1•Michelangelo11•9m ago•0 comments

Astronomers discover remarkable 3-star system just 28 light-years from Earth

https://phys.org/news/2026-11-astronomers-remarkable-triple-star-years.html
2•cybermango•9m ago•1 comments

Show HN: Greatai Review – free AI and human code review for open source

https://greatai.dev/clinic
1•alprimak•10m ago•0 comments

Universal Verification Methodology

https://en.wikipedia.org/wiki/Universal_Verification_Methodology
1•djoldman•10m ago•0 comments

Show HN: I built a site where companies can make offers to hire me

https://bidfor.dev/jamie
1•Jamie452•11m ago•0 comments

The GIF of a Bug Crawling on the Screen: Down the Rabbit Hole

https://veg.by/en/blog/2026/10/18/bug-gif/
1•vrubleg•11m ago•0 comments

Show HN: AI Inference for Beginners: 6 hours of video and 85 pages of notes [pdf]

https://github.com/samir-sarwar/samir-vllm/blob/main/InferenceFromScratch.pdf
1•samirswr•11m ago•1 comments

Show HN: ChronoStack-move time forward for your Docker Compose stack

https://github.com/Asmodeus14/ChronoStack
1•Asmo12•12m ago•0 comments

NASA's Perseverance, Odyssey Teams Investigate Technical Issues

https://science.nasa.gov/blogs/mars/2026/11/19/nasas-perseverance-odyssey-teams-investigate-techn...
2•ilamont•14m ago•0 comments

Magnus Carlsen 'Forfeit' Controversy at Total Chess World Championship

https://www.news18.com/sports/chess/explained-magnus-carlsen-forfeit-controversy-at-total-chess-w...
1•Tomte•16m ago•0 comments

Physicists Just Sent an Unblockable Message Encoded in Gravity–Using an Elevator

https://gizmodo.com/physicists-just-sent-an-unblockable-message-encoded-in-gravity-using-a-1950s-...
1•eric_h•17m ago•1 comments

Rent-Stabilized E Harlem Buildings Fail to Sell at Auction, Tenants in Lurch

https://citylimits.org/rent-stabilized-east-harlem-buildings-fail-to-sell-at-auction-leaving-tena...
1•gmays•18m ago•0 comments

Show HN: PermitCore – Sell desktop apps and games directly, licensed via 9 SDKs

https://permitcore.dev/
1•permitcore_dev•19m ago•0 comments

SSH Tunneling Has Unrealized Potential

https://probablymarcus.com/blocks/2026/11/20/ssh-tunneling-has-unrealized-potential.html
2•mrcslws•21m ago•0 comments

Double-Blind Submission Checklist for Researchers

https://presub.app/guides/double-blind-submission-checklist/
1•Gigacore•22m ago•0 comments

Claude Docs launch: 60 server errors across billions of requests

https://claude.dev/blog/chaos-loop-claude-docs-launch/
1•qainsights•22m ago•0 comments

Creative Agentic Computing Issue 2

https://codefarmer.com/magazine/issues/002/
1•artcube•23m ago•0 comments

Have You Tried Asking the People Who Do the Fucking Job?

https://abetterway.bearblog.dev/have-you-tried-asking-the-people-who-actually-do-the-fucking-job/
2•speckx•24m ago•1 comments
Open in hackernews

Show HN: AI Inference for Beginners: 6 hours of video and 85 pages of notes [pdf]

https://github.com/samir-sarwar/samir-vllm/blob/main/InferenceFromScratch.pdf
1•samirswr•11m ago
Tired of AI Slop learning? This took me months to write,record, and edit. I hope you can learn from this! It dives into a smaller version of vLLM and explains many deep learning concepts and optimizations from first principles! enjoy :)

Comments

samirswr•7m ago
I filmed over 6+ Hours of footage and hand wrote 85+ pages of notes to teach absolute beginners about AI Inference. Over the last 2 months, whilst working as a full-time SWE intern, I was grinding after hours to create this video series. I learned to code, amongst many other things, through YouTube videos aimed at beginners. Now, with the rise of AI, these videos tend to show up less, so I decided I wanted to change that. This is . A lot of thinking and consideration went into creating this, and all of my notes are written by me. For transparency almost all code was also written by me with exception of small bug-fixes, readme updates and stuff like that. There is also extra notes pertaining to each video which are AI generated, but main notes are all me. The series walks you through building a small Llama 3.2 1B inference engine in C++ and CUDA, starting with what happens when you send an LLM a prompt. I cover: How text becomes tokens, and tokens become embeddings. Loading model weights into GPU memory. CUDA basics, writing your own GPU kernels, and matrix multiplication with cuBLAS. The transformer step by step: RMSNorm, RoPE, attention, and the SwiGLU feed-forward network. Prefill and decode: processing your prompt and generating a response one token at a time. KV caching and how it saves repeated computation. Static and continuous batching to handle multiple requests. Paged KV caching and paged attention to manage and read cached data on the GPU. Here is the github repo as well as a link to my youtube: https://github.com/samir-sarwar/samir-vllm https://www.youtube.com/@samir-sarwar (if anyone hiring for summer 2027 interns focused on inference or mle hmu please!! )