Big Tech &
Intel selects Ohio for ‘largest silicon manufacturing location on the planet’ (3 minute read)
Intel will invest at least $20 billion on a new chip manufacturing site in New Albany, Ohio. The location will host two chip factories and employ at least 3,000 people. Intel expects the site to become the largest silicon manufacturing location on the planet, with plans to expand it to eight fabs. Ohio was chosen for the site as it was a better regulatory fit and building it would not displace any current residents. The site will become operational in 2025.
Leak reveals Microsoft’s canceled Windows OS for dual-screen devices (2 minute read)
Microsoft was working on a secret pocketable Surface device powered by a custom Windows OS before turning to Android for its Surface Duo. Andromeda OS had a lock screen that could act as a giant canvas to take notes, a home screen that worked as another inking canvas, gesture support, and a few other features that ended up in Windows 11. Microsoft was also working on a larger dual-screen device powered by Windows 10X, but that project was canceled along with Windows 10X. A 15-minute video showing Andromeda OS running on a Lumia 950 is available in the article.
Physicists have created “everlasting bubbles” (4 minute read)
In 2017, French physicists created spherical shells made up of plastic microspheres that could store pressurized gas in a tiny volume. These gas marbles could roll around on a solid surface without breaking apart. A recent study tested the longevity of the gas marbles and found that bubbles made from them could last between six and 60 minutes. Adding glycerol to the solution created gas marbles that could last from five weeks to 465 days. A lot of information about bubble science is available in the article.
Meta's 'data2vec' is the next step toward One Neural Network to Rule Them All (8 minute read)
data2vec is a multi-modal artificial intelligence algorithm from Meta that attempts to replicate the general learning ability that the human mind possesses. The team aimed to create an AI that could learn to perform many different tasks, including unfamiliar ones. data2vec can learn from speech, vision, and text without needing labeled training data. The technology will eventually be built into Meta's AR glasses to aid a virtual assistant. Details about how the neural network works are available in the article.
Programming, Design & Data Science
What Silicon Valley "Gets" about Software Engineers that Traditional Companies Do Not (11 minute read)
Silicon Valley companies consistently understand a few things that traditional companies fail to either understand or implement in practice. They allow engineers to be autonomous and figure out how to work, which enables the engineers to be problem solvers rather than just coding machines. There is a lot of transparency at Silicon Valley companies, with employees often having access to real-time business metrics, which exposes engineers to the rest of the business. Silicon Valley companies also encourage good communication between teams, pay more, and invest more into making the developer experience less frustrating. Engineers are treated as value generators in SV-like companies, rather than as factory workers.
Kedro (GitHub Repo)
Kedro is a Python framework for creating reproducible, maintainable, and modular data science code. It uses concepts like modularity, separation of concerns, and versioning and applies them to machine-learning code. Kedro features project templates, a data catalog, pipeline abstraction, coding standards, and flexible deployment.
This 22-year-old builds chips in his parents’ garage (11 minute read)
In August last year, 22-year-old Sam Zeloof produced a chip with 1,200 transistors using salvaged and homemade equipment. Zeloof created his first chip in high school as a senior in 2018 and was making individual transistors a year before that. While the chips are far underpowered compared to chips from chip makers with multi-million dollar budgets, Zeloof managed to progress his technology faster than the semiconductor industry did in its early days. Zeloof believes that making chipmaking more accessible to inventors would benefit society.
On Leaving Facebook (10 minute read)
This article talks about the experience of working for and leaving Facebook. Working on a new project at Facebook can feel like working in a small startup, but in a large company. Facebook makes it easy to switch between teams, and working there can be very comfortable. It aims to pay the top 5% in the market, which means that leaving will almost always mean a significant pay cut. All the benefits and compensation make it difficult to leave, but eventually many people get burnt out and start leaving anyway.