@vboykis @luketrailrunner Any nice training/inference discords you can share? Would love to be involved! Thanks!
English
Ananth Narayan
168 posts
Ananth Narayan
@Ananth2394
Todavía estoy aprendiendo | Machine Learning at Waymo
가입일 Ağustos 2011
274 팔로잉34 팔로워
@luketrailrunner A lot of discords for sure, especially in the ML training/inference/small LLM spaces, at proto community is also good, lots of building happening
English
I'm rebuilding AlphaFold2 from scratch in pure PyTorch.
No frameworks on top of PyTorch. No copy-paste from DeepMind's repo. Just nn.Linear, einsum, and the 60-page supplementary paper.
The project is called minAlphaFold2, inspired by Karpathy's minGPT. The idea is simple: AlphaFold2 is one of the most important neural networks ever built, and there should be a version of it that a single person can sit down and read end-to-end in an afternoon.
Where it stands today:
- ~3,500 lines across 9 modules
- Full forward pass works: input embedding → Evoformer → Structure Module → all-atom 3D coordinates
- Every loss function from the paper (FAPE, torsion angles, pLDDT, distogram, structural violations)
- Recycling, templates, extra MSA stack, ensemble averaging — all implemented
- 50 tests passing
- Every module maps 1-to-1 to a numbered algorithm in the AF2 supplement
The Structure Module was the most satisfying part to build. Invariant Point Attention is genuinely beautiful — it does attention in 3D space using local reference frames so the whole thing is SE(3)-equivariant, and the math fits in about 150 lines of PyTorch.
What's next:
- Build the data pipeline (PDB structures + MSA features)
- Write the training loop
- Train on a small set of proteins and see what happens
The repo is public. If you've ever wanted to understand how AlphaFold2 actually works at the level of individual tensor operations, this is meant for you.
Repo: github.com/ChrisHayduk/mi…
English
Computer Networking Fundamentals (mini-course)
Everyone who wants to understand container and Kubernetes networking should start from the basics. This heavily illustrated free course walks you through the first two layers of Computer Networking: Ethernet and IP.
A must for any developer or DevOps engineer who wants to build a solid understanding of computer networking before they move to the higher-level protocols like TCP, UDP, and even HTTP.
labs.iximiuz.com/courses/comput…
English
This is what I made by the way:
youtu.be/-Nvhuw_1iDM
YouTube
Vincent D. Warmerdam@fishnets88
This is a vibe-coded widget that I made to explore a differential equation. Good times!
English
Chrome used to have this lil Gemini icon on top right to click and ask questions of the page you're on. Did they kill that? I miss it tremendously. @GeminiApp
English
I teared up a little. Look at that baby go. First city outside of the US.
English
Looking for suggestions. What's a great computer science related book for a Christmas gift?
English
@archiexzzz @cHHillee jane street has really good technical videos. I remember watching one on GPUs
English
great learning hack: you open a really good deep technical video on YFLOPS by Jane Street @cHHillee -> parallely open google gemini's feature and topics you are not familiar with -> just use this feature to understand that. the best thing about this feature is that it asks you follow up questions on the topic you are trying to learn. like a class.
English
in the lectures below, i hold your hand through low-level LLM systems engineering. it includes everything up to TODAY!
1) pytorch tensors
2) large matmul on cpu vs gpu
3) JAX (and why xAI uses it instead of pytorch)
4) raw cuda kernels and global threading indexing
5) triton design philosophy and softmax example
6) HIP kernels
7) mapping out the ENTIRE ecosystem + differences between CUDA and ROCm/HIP (BLAS, FFT, DNN)
8) cutlass and cute-dsl
9) pretraining, finetuning, rl, unsloth, axolotl, megatron-lm, deepspeed, nanogpt, nanochat
10) training vs inference, inference serving problems, throughput vs latency vs concurrency scaling, vllm, sglang, tensorrt-llm, tensorrt, llama.cpp, exllamav2, exllamav3, benchmark comparisons
11) projects/companies using llms to generate SOTA cuda/triton kernels
12) luminal inference
13) mojo/modular/max
English
@charles_irl What's your karaoke song or choice? My go to is "If I ain't got you" by Alicia Keys
English
finally, a DD who won't complain when i sing karaoke at max volume on the way back from Napa
Waymo@Waymo
We're officially authorized to drive fully autonomously across more of the Golden State. Next stop: welcoming riders in San Diego in mid-2026! ☀️
English
*Guided lab sessions (finally!)*
As a book companion, I released several guided lab sessions on multiple topics (JAX, PyTorch, autodiff, mechint), more on the way, would love some feedback!
sscardapane.notion.site/Guided-lab-ses…
English
yesterday I walked 4 miles around Vancouver Canada and had a lobster roll for dinner
rly cool girl@hikingshawty
today i walked five miles and i had beef stew with dumplings for dinner <3
English
Career update: joined a US-based AI startup remotely, roughly $80K USD (including stock) 🙏
English
It’s easy to tell who in SF is getting laid and making money and who is here for YC F25
English
@not_by_default @yoeven @supabase @patrickc @ycombinator Yeah brother gonna have tree trunk legs as a unicorn 🦄
English
English
@not_by_default and @ritanshu675 have immense technical expertise, and understand the problem space well.
Questom is fresh out of Y Combinator, and they also have a lot of customer love and a very innovative product. If you’re in apparel or ecom, keep an eye on this one. 🚀
Abhimanyu Yadav@not_by_default
This is not the post we had planned. In fact, we locked in the content of this post last night in a conversation at 11 PM, and I’m doing @ritanshu675 dirty by rewriting it last minute. I know he will forgive me, he usually does. In that same conversation last night, Ritanshu, a man of few words, asked me a simple but profound question: “What are the things that bring you joy?” I launched into a long list - the tight hug from my wife every morning, watching my nephew grow into a kind and loving human, hours at the gym with my best friend, the small chores that quietly make a house feel more like home, the daily voice memo from my mom that begins, “Om Shanti Abhimanyu.” On and on I went, until I finally stopped and asked, “Why do you ask?” He smiled and said: “Because we already have everything that brings us joy - everything else is a blessing.” And with that sense of gratitude for the lives we already get to live, we are overjoyed to share that we’ve been accepted into @ycombinator . We want to represent a different archetype of founder - one rooted in stability and persistence. Yes, sometimes we’ll pull an all-nighter or catch the flight that leaves in two hours to make it to conference we just learnt about. But just as often, we’ll come home, fade into the Inner Sunset mist, and find joy in the everyday. OK, now that you know us and what we stand for, let us share what we are building. We’re quietly working on something small, that could be great — AI Sales & Support for the commercial printing industry. You know, the invisible force behind fundraiser t-shirts, business cards at career fairs, and the billboards you pass on your commute. Today, too many hours in this industry are swallowed by chasing specs and waiting on approvals from customers. Meanwhile, good leads slip away. We’re building an assistant that never sleeps: qualifying, following up, and keeping the pipeline warm - so the teams can spend their hours where they matter most. And here’s where the beauty of the internet comes in, where you come in - if you know anyone at @HelloMinted , @Vistaprint , @customink, @Shutterfly , @Printful, @StaplesStores tag them here, or make an introduction. Help us turn this little win into something great. 🌱 Check us out at: questom.ai
English