Gergely Orosz

Two years ago, Figma was valued at $2B. Today, announced selling for $20B. This makes Figma valued more than these public tech companies:

Box $3.8B

Dropbox $8B

Etsy $14B

Twilio $14B

Pinterest $16B

Coinbase $17B

Expedia $17B

Snap: $19.4B

Spotify $19.6B

Incredible outcome for the Figma team, but also likely a major win for Adobe in buying the company that could have disrupted them. It also shows that going public is not always the most profitable exit.

#adobe #figma #startups

---
Follow me for content on software engineering at Big Tech and startups. Subscribe to The Pragmatic Engineer at https://lnkd.in/grXSBkVw for deep dives on these topics.

The definition of what “region“ in the cloud is different at AWS, and at Google Cloud and Azure. Why is this important? Well, one incident in one data center took a whole GCP *region* offline!

When I researched differences between cloud providers, something stood out: AWS treats their region definition much, much more strict than other cloud providers do.

At AWS, a region consist of 3 or more zones (one zone can be 1 or more data centers). Zones are at least 30 miles away from each other and usually no more than 60 miles, to mitigate against local disasters like earthquakes, floods, extreme weather etc. This means that if a data center has issues, or even if a whole zone goes offline, the region continues to operate. So customers who operate in two or more zones within a region can keep operating in this region.

On the other hand, Google Cloud - as well as Azure - is a lot more vague on what a region is. Sure, they both say how a region consists of multiple availability zones: but neither is clear on how “separate“ a zone is. Neither guarantee geographic separation, or even building separation. Both make it clear that zones *are* independent of one another for e.g. power supply, networking etc. But no more specifics than this are given.

Normally, the difference is subtle, as zones have some level of separation. Except when a local disaster strikes!

Google Cloud has water intrusion in one of their data centers, impacting one zone, in eu-west-9, which is in Paris. What this would normally mean is one zone is impacted.

However, in this case, the whole region - with several zones - went offline.

How could this happen?

Is the whole region in one building, without geographic separation and so could have “water intrusion“ hit the whole building? Or was it a critical service operating in the zone impacted, took down the whole region? If so, why were zones inter-connected so tightly?

We don’t have answers to either of these questions just yet.

One thing is certain: “region“ means something different in AWS than in other cloud providers. Incidentally, both Google Cloud and Azure claim to have more regions and zones than AWS, globally. But the comparison is not entirely apples to apples, once you realize how some cloud providers locate a region in one location (or building), while others spread them out, to allow for much more reilliency for disasters that can happen in a data center.

Google will have to provide a detailed postmortem to restore trust, and explain how a problem in one data center can take a whole region down, and what improvements they will do to avoid this in the future. More details on the GCP outage: https://lnkd.in/emyBJke2

If you're a decision maker on choosing cloud providers, pay attention to the details on how they define a “region“ or a “zone!“

---
Follow me for interesting topics on software engineering at Big Tech and startups. Subscribe to The Pragmatic Engineer at https://lnkd.in/grXSBkVw for deep dives on these.

Both Uber and Doordash have removed the 1-year equity vesting cliff for employees: and this is a BIG deal.

Until now, a 1-year cliff was the norm in most of big tech**. If you left - or got let go - before a year, you saw nothing of your equity. This was especially an issue when being laid off, especially for no fault of your own, like it happened around COVID. Employees 7-9 months in saw none of the equity that was a meaningful part of their compensation, which was both unfair and frustrating.

I am hearing the change came after hiring - especially software engineers - is becoming increasingly heated and slowing down across most markets. Not having a 1-year cliff makes a whole lot of a difference as an employee - giving more compensation, earlier - while it basically costs the company little to nothing extra.

I am expecting we’ll see more of big tech follow. Kudos to Uber and Doordash for making this employee-friendly step!

** Update: people have let me know Google and Facebook removed the cliff 2 years ago, Pinterest 18 months ago and Spotify and Snap also no longer have it. All the more reason for other publicly traded big tech to follow.

GOOD NEWS: plenty of companies are still hiring software engineers and engineering managers. Here is a list of more than 1,300 companies that are doing so. Check them out:

More than a month ago, I set out to collect a list of companies still hiring software engineers and engineering managers. The first iteration of this list was at 750 companies, and this grew, over the weeks.

The list is based on submissions, and I've now closed adding more companies to it, after gathering - and reviewing - submissions for weeks.

Of course, these are not the only companies hiring. They are the ones where at least one person working at the company submitted details on how they are, and added a few sentences on why they think it's a good place to work at. So it's a good starting point for anyone looking around on the market.

See the full list here:
https://lnkd.in/eed-Fibj

If you're planning to apply to these places, I wrote a book with advice on software engineering resumes: https://thetechresume.com/ It's free for anyone currently without a job: https://lnkd.in/gfBqxMKu

#hiring #softwareengineering #engineeringmanagement

---
I write about software engineering at Big Tech and startups. Subscribe to The Pragmatic Engineer at https://lnkd.in/grXSBkVw for deep dives on these topics.

If you are living in Europe, Putin's war on Ukraine is about you as well. Putin's open goal is to restore the USSR territory and influence, penetrating into central Europe over the next decade. If Ukraine falls, he will not stop: he will get even more bold.

I was born into the Eastern Block and my family had the chance to settle in the US in 1990, getting a green card. We chose not to, coming back to Hungary because my parents saw a future that did not involve being occupied by Russia and one without the Cold War and the threat of a nuclear standoff.

Putin now wants to bring back of what he thinks is the glory days of the USSR and the Eastern Block. When Hungary, Poland, Romania, Bulgaria, Czechia, Slovakia, Ukraine, Latvia, Estonia, Lithuania, Albania were all under Russian influence. Citizens lived in a world of quiet fear, one where teaching Russian was mandatory in schools, fleeing West close to impossible and where Russians could enter with tanks when they pleased - and they did! It was a time where the though of living freely in a democracy was but a dream.

I don't want this past to return. Right now, Ukraine is the only thing standing in Putin's way on this road.

I #standwithukraine, not just with words, but using the means that I have. Starting now, my family and I are redirecting what we have put aside as monthly savings to supporting refugees and the Ukranian army.

If you are European: the future of Europe is being decided now. I urge you to act, so this war is not just Russia vs Ukraine, but Russia vs Ukraine supported by all of Europe.

#europe #future

How does a migration of 1 million lines of code, from a Python monolith, to Go services look like? In the case of Khan Academy it looked like this, spanning 3.5 years:\n\nI talked with Brian Genisio and Kevin Dangoor - both of who played key roles in this migration - to learn how the company went through this exercise of not only changing languages (Python -> Go), but changing APIs (REST -> GraphQL) and architecture (monolith -> GraphQL federation with smaller services).\n\nThe team started by agreeing on the architecture approach, and then migrating the smallest and simplest possible service to this new architecture, and sending production traffic onto it. This was a service that answered the question of what was the oldest version of the mobile client that could be supported, and accounted for probably less than 0.01% of website traffic.\n\nThe team then diligently migrated field after field, endpoint after endpoint.\n\nAt the halfway of the project, about 95% of traffic was migrated over. However, this was the point when the existing tools needed to be rebuilt - or re-fitted - and the services needed new capabilities, and some new features to be added.\n\nStill, 3.5 years is a long time, and, as Brian shared:\n\n“By the end of the project, all of us: engineering, product, design and the rest of the business, just wanted the migration to be over so we could move into the post-migration world.“\n\nRead more details on this migration on this week's free The Pragmatic Engineer issue here: https://lnkd.in/egsKTWzJ\n\nAnd thanks to Brian and Kevin for sharing details and learnings!\n\n#softwareengineering #migration \n\n---\nFollow me for content on software engineering at Big Tech and startups. Subscribe to The Pragmatic Engineer at https://lnkd.in/grXSBkVw for deep dives on these topics.

Scoop: Uber is closing its engineering office in Vilnius, Lithuania. The decision impacts about 60 software engineers. The decision was communicated in what is a considerate manner, giving ample heads up.

I knew some of the engineers from Vilnius from my time working at Uber, and meeting them on events. The site always had a heavy infra and developer tooling focus. Teams on this site own or co-own infrastructure offerings like Cassandra, Elastisearch, container deployment/management, OS configuration, oncall tooling. A project called Devpods - a remote developer environment similar to Gitpod - has been built in Vilnius and is fully owned here (originally started in Amsterdam in 2020, thanks for the correction Chris Laffra!)

On Wednesday, 7th of September, Uber’s leadership announced the Vilnius site would be permanently closed. The directors of their respective teams came to the office to share the sad news that Vilnius, as an engineering site, will be closed down. The reason? Lack of local, senior leadership.

Having talked to software engineers at Uber, in Vilnius, the reasoning did have merit. Hiring managers was challenging in this location, and there have been influential managers in the past who moved back to being an individual contributor.

Uber is handling the closing of the office in a considerate way. There are no layoffs on the spot, and nothing is rushed. The site will close in June 2023 - almost a year away. Uber will support people who want to relocate and move to other offices of their choice. The performance review process will run its course in early 2023, and bonuses will be paid around March 2023.

While closing an office is never good news, it's nice to hear that Uber leadership is not rushing this process, and all directors made the effort - and the trip - to share this difficult news in-person.

If you are hiring for outstanding engineers, I'm sure some of the folks will be open to new opportunities: either now, or in a few months. I had nothing but great experiences with the people I worked with.

Sending support to Uber Engineering in Vilnius and a special thanks to those who built - or worked with - the in-house oncall tooling that I opened almost every day when I still worked at Uber. A great team 🔥

I covered more about this event in this week's The Pragmatic Engineer: https://lnkd.in/g9ngZmBD

#uber #lithuania #softwareengineering

---
Follow me for content on software engineering at Big Tech and startups. Subscribe to The Pragmatic Engineer at https://lnkd.in/grXSBkVw for deep dives on these topics.

LinkedIn: what the hell? How is it ethical (or legal?) to use “messages“ to push ads from fake profiles, and with no way to opt out? Even as a paying customer??

Recently, I started to see ads - disguised as messages - in my LinkedIn account. They do not come from a real person: they seem to be AI-generated (you cannot respond to them, and the “profile“ does not exist - it cannot be clicked or responded to. For legit messages sent by a human you can reply and click through to their profile)

I happen to already be a paying LinkedIn member, paying around $600/year for keeping my messages open, and for the occasional InMail.

LinkedIn is truly losing its way, and trying to milk every last cent, integrity and ethics be damned.

Who in the LinkedIn product team thought this is acceptable, or that fake/AI profiles are on-brand for a supposedly trusted platform?

PS: I've now cancelled my LinkedIn Premium subscription. To message me on other platforms, for free, please see https://lnkd.in/eR2Dfg9f (I paid for LinkedIn Premium for anyone to be able to message me - but given how DMs can no longer be trusted to come from humans, I see no reason to encourage anyone to use LinkedIn DMs. Message me on email, Bluesky, X and Signal: all platforms where a DM means it is coming from a person, and a not bot built by the platform, trying to get more ad views.)

PPS: LinkedIn might be the only platform spamming paying customers more than free members. See the comment from Michael Malone below:

“Ever since I cancelled Premium, I have received far less spam. LinkedIn might be the only platform where you actually see more junk by paying for the service.“

Something amusing I heard recently: a Fortune 500, non-tech company is NOT allowing devs to purchase one of the popular AI coding tools (e.g Cursor, Windsurf, GH Copilot, Augment Code etc).

Instead: they are building their own, internally.

The official reasoning: IP, security, compliance.

The unofficial reason (that no one will say, of course): getting headcount for a team; playing around with LLMs, and a sure and trusted path to get promoted by shipping this project and demonstrating company-wide impact.

It's amusing because now this company has subpar coding tools compared to their competition that allows devs to get tools built by startups specializing in dev tooling - and all of this for a higher price in the end!

And we can see how this will play out already: the project ships, gets some usage, people get promoted, then in 6-18 months the company will finally sign an enterprise contract for a dedicated AI coding tool that works better, more reliably and has far superior developer experience.

We've seen this in the past with e.g. “private cloud“, Big Data solutions, CRMs or even chat tools (“Slack would not scale for our use case“ - then four years after building the internal chat tool that was subpar in features and reliability, the company finally moved over to Slack and turned out Slack 'scaled' just fine for tens of thousands of people).

Looks like this time is different - but some things remain the same!

Teams with a high quality bar for code reviews are great places to grow professionally. But this same high quality bar can feel intimidating when being a recent joiner.

Empathetic engineers pay additional attention for new joiners when doing the first few code reviews:

(The below is from the draft of The Software Engineer's Guidebook I'm currently wrapping up, as it will go into the content review phase https://lnkd.in/gq2722ZF. If you are interested in giving feedback on the draft contents, you can indicate this here: https://lnkd.in/eHWGf_QK - and thanks in advance!)

#softwareengineering #codereview