Really great article from Optiver. It expanded my view of what Postgres is capable of, especially using it to track data changes in real time without adding extra tools. Nice to see Postgres pushed beyond the usual.
optiver.com/insights/technol…
Fun read from GitHub eng blog.
The Stargazers API was initially designed for one purpose but largely used for something else. Interesting to see how they listened and built for this informal use case, making the platform better and the community happy.
github.blog/engineering/arch…
Thursday was my last day at Univrses.
4+ years, big milestones, and the best team I could ask for. The high-trust culture taught me what real product ownership looks like.
Grateful to everyone I worked with. Leaving was tough, but I’m excited to share what’s next soon 🥳
Don’t just put in the work, own the work!
Really great article, please read.
I sense chaos ahead. The software development lifecycle is more concentrated than ever. Only a handful of companies are building foundational models, but huge and expanding ecosystems of layers are built around them, finding their way into almost everything...
I needed to generate custom receipts over the past couple of months but couldn't find a tool that worked exactly how I wanted, so I built one.
Simple, browser-based, no sign-ups. Give it a try if you ever need something similar: iamotz.github.io/receipt-bui…
When implementing the resend option in a passwordless auth setup (magic link or OTP), should the system resend the same link or code if it's yet to expire, or always generate a new one?
I'm trying to think through this, and I'm not sure which is better 🤔
TIL JavaScript lets you use numeric separators (underscores). You can write one million as 1_000_000 instead of 100000. Same number, way more readable, and it just works out of the box, superb! 🤩
At its core, anything that lets you store & retrieve data is a database. A variable, a file - all qualify. But the moment you care about *how fast*, *how long*, and *how much* data you're dealing with, the data structure underneath starts to matter a LOT.
2. LSM-Trees - fascinating architecture. Writes go into a sorted in-memory buffer, flushed to immutable sorted files (SSTables) on disk. Naturally biased toward HIGH write throughput. Cassandra & ScyllaDB are built on this.






