puang

we built something that makes your codebase talk back @devdoq auto-docs your entire project while you build no extra steps, no context switching, just clean and searchable docs log in at app.devdoq.com/login to join the waitlist ;)

Yesterday: the most dangerous line in fintech. Today: what happens when you try to fix it. You wrap it in a transaction. You add SELECT FOR UPDATE. You sort the IDs to avoid deadlocks. You add idempotency. You add retries. Now your "simple balance update" is 150+ lines of custom locking code that only one person on the team understands. And it still breaks under load.

in @heliumbrowser is there a way to get this tab switch feature? this is the only thing im missing from my arc experience

ux of @heliumbrowser improved significantly, now i can make the switch finally

yooo no way matlab??

@sharpeye_wnl cfbr

am looking for ML/AI intern opportunities , open to both research and applied roles about me : I'm a sophomore student interested in machine learning and ai related systems. I have been working in this field for a while and here are my projects: github : github.com/sharpeye08 > kakugen (in progress) : character-level LSTM trained from scratch on a curated philosophical corpus to generate aphoristic prose. Built custom training loop with perplexity tracking, temperature sampling, and multi-source corpus pipeline in PyTorch. > anzen : Real-time road safety alert system using YOLOv8 to detect traffic hazards and fire severity-graded alerts. > kenkyū : Autonomous AI research agent built with LangGraph for multi-step reasoning and web search, running fully locally. > ResNet18 Transfer Learning : Fine-tuned ResNet18 on CIFAR-10 via transfer learning, hitting 93%+ accuracy with a from-scratch training loop. > Sharpner : a no code AutoML platform where users upload datasets and train ML models through a Streamlit UI. you can check more of my work at sharpeye08.github.io

@AdityaMandal_ cracked!!

Built my own implementation of Kafka Paper in Rust [repo in replies] the implementation is focused around storage and replication model, it covers: - append only segmented partition logs - sparse offset indexes - CRC protected record batches - batched produce and batched fetch - pull based consumers with offset replay - time based retention - paper style consumer groups with deterministic rebalance - an explicit coordination plane for brokers, metadata, claims, and committed offsets - multi broker topic placement and leader metadata - leader/follower replication with watermark based visibility - failover, leader reassignment, and broker restart recovery this project was fun to build, Readme file covers more things in detail Do check it out

@Disha16twt cfbr

Looking for ml, backend related internships, any leads or feedback is appreciated :)

was getting bored so wrote this simple flashcard feature for lexit now it is much more easier to remember meaning of words and enhance your vocab :)

@aetosdios_ cool work man!!

Most image viewers still carry the wrong assumptions. > Iris is being built the other way: > GPU-native > Linux-first > zero-copy where the stack allows it > raw Vulkan at the renderer Small first cohort soon. Link to waitlist in comments

open.spotify.com/album/5poA9SAx… #ye

@aayushman2703 🧌🧌

@notpuang Answering questions like this

what are some problems that you face daily i'll try to solve it

@shbhtngpl brehh 😭🙏

@notpuang my mind is the problem, please solve

was going through emilkowalski's blog on animation tips and came across this cool button so i recreated it using pure html css

was going through @emilkowalski 's blog about animation tips and came across a really cool button animation so i tried to recreate it with pure html css btw

@mihirtwt oh interesting, will look into it

@notpuang damn, u can even add a shader generate one from unicorn studio

experimenting with the cta section on the landing page,