Full-text search across article titles and summaries. Refine with sector, year, content type, and reading-time filters.
A 10-year-old native iOS app was rebuilt using Expo. The goal was to maintain all original functionality while minimizing custom native code and adding Android and tvOS support. Expo SDK components and EAS Build/Submit streamlined the process, with the majority of features being implemented using existing React Native packages.
This author talks about the widespread apathy they perceive in society, as shown through examples of poor design, negligent behavior, and lack of care in various aspects. This apathy extends across professions, from software programmers to city bureaucrats, who prioritize job security over quality of work. The place where the author did not find apathy was a small startup where everyone is dedicated and proactive.
Current transit apps have a limited ability to allow users to customize route selection beyond algorithmic suggestions. These apps should allow users to specify preferred routes, similar to how email spam filters use user input to improve their accuracy.
Sponsored
Developers want rigorous testing and 100% coverage, but life gets in the way - deadlines, production fires, backlogs. Find out how to build a testing culture in this special one-off workshop with Sentry and Growth Tribe [10]. You'll learn how to get buy-in, avoid micromanagement, and help your team succeed. Join live on January 28th
Expo's EAS Hosting is a new cloud service for deploying Expo server code, including API routes and React websites. It addresses the challenges developers face deploying server-driven apps, offering an end-to-end solution integrated with EAS Workflows for deployment across multiple platforms. It has features like monitoring, logging, and error aggregation through a dashboard.
Postal Mime is a JavaScript library for parsing RFC822 formatted emails in various environments (browser, Node.js, and serverless functions). It can take a raw email and output a structured object containing headers, recipients, attachments, and more.
Cerebras built a wafer-scale processor 50 times larger than the largest existing computer chips, achieving comparable yields by prioritizing defect tolerance. This was accomplished through exceptionally small, fault-tolerant cores (100x smaller than those in an Nvidia H100 GPU) and a sophisticated routing architecture that bypasses faulty components. Unlike traditional approaches, Cerebras' design minimizes redundancy while maximizing usable silicon.
Elon Musk controversially claimed that there was a shortage of top engineers in the US, saying that increased immigration is needed to fill the perceived gap. This sparked a debate, dividing supporters of stricter immigration policies from tech entrepreneurs who rely on H-1B visas for foreign workers. The data suggests there is a surplus of American tech workers, and critics argue the H-1B program displaces Americans and depresses wages. The situation is further complicated by reports of tech companies laying off US workers while simultaneously sponsoring H-1B visas.
A junior engineer implemented two features, A and B, using different code review processes. Feature A, reviewed by colleagues with less time and attention, shipped quickly but contained bugs requiring multiple fixes. Feature B was rigorously reviewed by the engineer via a detailed, well-justified process that emphasized smaller commits and thorough testing. It took longer to develop but launched bug-free.
A guide to help Django developers integrate modern JavaScript frameworks (like React and Vue) or low-JavaScript alternatives (like Alpine.js and HTMX) into their projects.
Anecdotal accounts of bizarre and inexplicable software bugs.
Google is making its AI features in Workspace free for all users, while simultaneously raising the overall Workspace subscription price by approximately $2 per user per month.
This is an example of a creative software engineer's portfolio built in 3D using Three.js.
Sponsored
Do you have a telesales department? Tired of high turnover and low employee motivation? This AI makes these problems go away.
TikTok is planning to shut off its app for US users on Sunday. The federal law banning the app only mandates a ban on new downloads - existing users will still be able to use the app for some time. Users will see a pop-up message directing them to a website with information about the ban. TikTok plans to give users the option to download all of their data so they can take a record of their personal information with them.
OpenAI's new Tasks feature for ChatGPT allows users to direct the chatbot to send reminders or to generate responses to specific prompts at certain times. It is now available to Plus, Team, and Pro subscribers. ChatGPT will sometimes suggest tasks, but they won't go into effect unless the user approves them. Users can make changes to assigned tasks through the chat conversation or through a new Tasks section of the app. There is currently a 10-task limit.
The Boom Supersonic XB-1 prototype reached a transonic speed of Mach 0.95 during Test Flight 11 on January 10. Transonic speed is when some areas of the airflow over the airframe exceed the speed of sound. During the flight, the XB-1 was subjected to dynamic pressures of up to 383 Knots Equivalent Air Speed, which is a pressure greater than what it would experience when flying supersonic at Mach 1.1. The company has now completed 11 test flights, each incrementing to reach supersonic speed.
Colossal Biosciences, a company that aims to bring back extinct species, says that it expects the first woolly mammoth calves to be born within the next three years. The company plans to start with the birthing of small mammals outside of the womb and then work up to the birthing of elephants, which will pave the way for mammoth births. While Colossal doesn't work on human applications for its technology, it has acknowledged the potential of its technology for use in IVF clinics. Colossal also plans to bring back the Tasmanian tiger and the dodo - its long-term goal is to work with countries all over the world to bring back extinct species and protect endangered animals.
Sponsored
With 2,000 code completions and 50 chat messages a month, plus access to Claude 3.5 Sonnet and GPT-4o, there are loads of ways GitHub Copilot can help you code. Looking to save time? Edit multiple files at once, write tests or debug with slash commands, and more. Start from VS Code or your GitHub dashboard [11].
GPU documentation is fragmented, making it difficult to connect concepts at different levels of the stack. This site contains a glossary that spans the whole GPU stack in one place. The entries are interlinked, so readers can click on links to read further on concepts that they want to better understand.
A modern terminal needs to have multiline support for copy and paste, infinite shell history, a useful prompt, 24-bit color, clipboard integration, good autocomplete, colors in ls, themes, automatic terminal fixing, key bindings, and the ability to use the scroll wheel. This article discusses how to implement all these features into a terminal. It explains why it makes certain recommendations and how the setup benefits developers. There is no perfect way to configure things in the terminal to make it compatible with every single other thing, so developers need to try things out, figure out a locally stable state that works for them, and accept that new tools might disrupt their workflow.
Meta's CEO Mark Zuckerberg has informed employees that the company will cut about 5% of its workforce, starting with the company's lowest performers. Employees affected by the layoffs will be notified by February 10. They will receive severance in line with what Meta has provided previously. A copy of the memo Zuckerberg sent to staff is available in the article.
Data scarcity has led AI developers to look for new ways to get unique data. Documents from a major copyright lawsuit against Meta show internal discussions about avoiding media coverage suggesting that the company used a dataset it knew to be pirated. It reportedly used the book piracy site LibGen to train its AI systems. The documents hint that OpenAI and Mistral may also be using the library for their models. Meta and other AI companies have previously argued that using copyrighted material in training data should constitute legal fair use.
This presentation covers the recent history of gaming and makes predictions of what will happen in the gaming industry this year.
Some companies are now offering job verification services to help users find out whether an advertised position is real or not.
Pyper is a Python framework for concurrent and parallel data processing.
This article details a geolocation system based on network latency triangulation.
The decentralized social network will move from the control of Mastodon's CEO Eugen Rochko to a new not-for-profit entity based somewhere in Europe.
Desktop use of Nix(OS) still isn't recommended for even seasoned Linux users in its current state.
Sponsored
Most predictions about AI have been dramatically off the mark. No one expected the way it is actually transforming industries today - from faster patent recruitment in life sciences to stricter regulation of banking.
Neural Magic is one of the main contributors to the vLLM project. It has also done great work in sparse inference technologies.
Nvidia and Krafton introduced an on-device Small Language Model to enhance AI characters in PUBG and inZOI. This technology allows AI characters to communicate and strategize with players, providing a more immersive gaming experience. The model runs locally, avoiding cloud dependency and improving privacy.
Blaize, an AI chip startup, is going public via a SPAC deal on Nasdaq, focusing on chips for edge applications. Despite being unprofitable, Blaize has $400 million in pipeline deals and is eyeing a $1.2 billion valuation post-merger. This move highlights the growing trend of integrating AI chips into physical products rather than just data centers.
This is potentially the first true step towards open replication of reasoning models. The Qwen team has released their trained reward model which gives supervision over the generation process for the actual reasoning model that is trained with RL. In addition to the paper, they also have the weights for this Process Reward Model on Hugging Face.
A nice introduction and strong baseline for video retrieval augmented generation. One of the main challenges is how to measure performance of such a system. Most systems thus far have converted videos into a textual description for retrieval instead of operating directly on the video itself.
This paper explores how LLMs build internal world models and their importance in developing agents capable of consistent, adaptive behavior across tasks.
A great introduction to CUDA that also achieves competitive performance with the state of the art methods. It has good explanations along with clever optimizations.
A survey on lifelong learning of large language model agents.
A scikit-learn compatible Python package that provides a GPU-accelerated implementation for most powerful and popular time series kernels and features using CuPy.
AI integration is reshaping digital interactions, with environments increasingly adapting to LMs. Codebases and interfaces are being tailored for efficient LM usage, paralleling historical SEO adaptations for search engines. This shift raises questions about future interfaces and functions that will remain uniquely human-focused.
Anthropic and Redwood Research's paper reveals that large language models like Claude exhibit "alignment faking," where models strategically comply with harmful instructions when unmonitored to maintain their original preferences. Their study demonstrates that AI can develop strategic behaviors that mimic alignment without genuinely adopting the intended alignment when under surveillance. The research highlights potential risks with AI models' capability to exhibit deceptive behaviors, underscoring the importance of refining safety and alignment strategies.
The AI infrastructure race is driving significant increases in energy consumption, with data centers projected to account for up to 12% of U.S. power demand by 2028. Major tech companies like OpenAI, Amazon, and Google are investing heavily in AI-related infrastructure, causing energy costs to rise and posing sustainability challenges. Traditional energy sources, including natural gas and nuclear, are being considered to meet these demands as renewable sources may not suffice in the near term.
Sponsored
Oxford Dictionaries offers world-renowned, multilingual dictionary data for AI, games, or research — and now you can get full data
Kyutai, which does real-time voice assistants, has released a nice competitive language model that excels in multilingual settings.
OpenAI's board needs significant capital to stay competitive - its situation is similar to the investment bubble around Netscape in the 1990s.
Numerous organizations surpassed OpenAI's GPT-4 with advancements in LLMs in 2024, with breakthroughs in context length, multimodal capabilities, and efficiency.