Gergely Orosz
@GergelyOrosz
Writing @Pragmatic_Eng, the #1 technology newsletter on Substack. Author of @EngGuidebook. Formerly Uber & Skype.
Substack launched analytics insights and: wow. Thank you to everyone reading @Pragmatic_Eng!
A very similar automatic systemd update on Ubuntu 22.04 caused the biggest-ever Datadog outage in March 2023. In June 2025 it caused the biggest-ever Heroku outage. Deja vu... More in today's deepdive: newsletter.pragmaticengineer.com/p/why-reliabil…
So @GergelyOrosz has a newsletter out on reliability at scale (and why it's so hard), with lessons from Heroku, GCP and Neon. Go read it! substack.com/@pragmaticengi…
This comp story is in India: 40 lakh is ~$46K per year or ~€40K per year. I would think tech companies need to pay a premium if asking for 5 days in-office. Anyone who doesn’t do so: you’ll likely only be able to hire candidates who get no other offers, or prefer daily commute
Got a call from Zepto HR Asked my expected base, I said 40 She said max we can give is 38 At this point I had interviews going on, so I was like okay sure Then she said, "You’ll have to come to the office every day, and only if it’s some emergency and you have to go home,…
It is striking to see how good obsessed Heroku was about reliability in 2010, and how the Heroku in 2025 has degraded so bad. (I was looking at how incidents were handled back in 2010, vs now) Is this the "Salesforce effect?" Hard to unsee the decay in this area.