Manuel Klimek (@manuelklimek) profile picture

Software Creator; he/him

Munich
Joined February 2009
Photo by Manuel Klimek (@manuelklimek)Photo by Manuel Klimek (@manuelklimek)Photo by Manuel Klimek (@manuelklimek)Photo by Manuel Klimek (@manuelklimek)Photo by Manuel Klimek (@manuelklimek)Photo by Manuel Klimek (@manuelklimek)Photo by Manuel Klimek (@manuelklimek)Photo by Manuel Klimek (@manuelklimek)Photo by Manuel Klimek (@manuelklimek)
new little song: "The Re-Org Rag (I'm My Own VP)"
218
2,623
817
10,100
Playing 🐒 🏝️ I realize how much like leadership in big tech it is: VPs give you a cryptic quest. You run around talking to people until you find the right talking points; talking to one person gives you new options with another one. Your inventory is full of documents.
2
12
I don't care about math!! I know too much about programming language design to ever want 84 - 72 / 6 to equal 6!!
1
We had quite some fun for the last 2.5 years fuzzing CPUs. We wrote one system, scratched it, and wrote another one. This week we open-sourced most of it, and hope to open-source more in the future. github.com/google/silifuzz
5
132
4
596
Thrilled to announce the 3rd cohort of @GoogleStartups Black Founders Fund! Founders receive $100K in non-dilutive funding, $100K in cloud credits,free therapy & PR support. We're also sharing the next 7 Black-led funds to receive investment from Alphabet. blog.google/outreach-initiat…
12
35
16
235
The old problem: trust is incredibly easy to lose. Automated solutions need to be *better* than humans. Our teams at Google do automated changes on >1BLOC of code. We have a team reviewing those changes ahead of time to make sure they're top quality.
This Post is from an account that no longer exists. Learn more
1
4
@TitusWinters @kevinb9n @chandlerc1024 at some point we need an OSS "large scale change" committee :D
5
Change cost scales with organizational size. Combined with a long tail, that means that break-even for fundamentally good changes in large organizations can be a decade+ (e.g. changing programming languages). Getting people to plan past their retirement goals is hard.
1
6
While I was actively doing SWE interviews I was ~always able to say "I used a similar algorithm in the past month". If you analyze the behavior of large C++ programs / builds, it helps to be able to quickly write an efficient graph algo. That said, I'd never use "insights".
“So it says here you’ve maintained steady employment for 15 years as senior software engineer. Alright cool, we’re just gonna have you solve some algorithm stuff that you, and honesty we, have never actually used at work”
1
1
My go-to was something that required recursion and a bit of state. All that said, I'm aware I was still unable to distinguish "can't do it" from "can't do it under pressure in high stress situation". As an industry, widely implementing a better solution to that would be *huge*.
1
OSS development is inefficient (velocity wise) because time to feedback for anything that's not a leaf project is way too long. Some ecosystems are faster than others, but TTF should be minutes or at most hours. Are there any large software ecosystem that truly live at head?
1
5
Manuel Klimek retweeted
Four reasons why I think authorship on academic papers should be more like movie credits:
42
742
128
4,837
Since I've become a manager I've got better and better at sheets. Finally less need to bring appscript in.
🚨🚨 BREAKING: 🚨🚨 If you're a Google Sheets user, stop what you're doing and check this out... ...they just dropped 10 new functions on us!!! 🔥🔥🔥 Christmas came early for Sheets nerds 🎅 Let's check them out 👇
2
So much this. Teaching my now 4.5yo stuff, the best first <thing> is always the one that motivates you to learn.
This Post is from an account that no longer exists. Learn more
1
Manuel Klimek retweeted
It shouldn't be that simple to have the top linux, git and curl developer show up in a fake project. Just a test, but a disaster waiting to happen with blind trust. Need scalable solutions - signed/verified commits, trusted identities, etc- github.com/azeemshaikh38/ten…
1
3
15
Manuel Klimek retweeted
Proof that fuzzing can discover exploitable vulnerabilities that aren't memory corruption! OSS-Fuzz discovered a very interesting command injection vulnerability which was just fixed: github.com/syoyo/tinygltf/is…
4
26
4
95
I've indulged in snark, sarcasm, being bitter and mean about things I didn't like, and I want to say... It wasn't worth it. I regret it. Snark leads to anger... [Fill in Yoda's quote] It always ends toxic. Life’s too short to be snarky all the time. It’s just not worth it.
10
25
181
Automated testing is not about bugs, but about productivity. For business success, productivity (within limits) doesn't matter that much. Building the right thing at the right time matters. That's why product management is such a high level bit.
As a software engineer, when you hear about a practice that doesn’t match your quality “values”, the first instinct is to dismiss it. But it’s just as worthwhile to understand why these practices can work. Eg how FB did few tests, but invested in automated stages rollout layers.
1
Data is hard. This is why science takes years until there is real consensus on facts.
This meme might not be as simple as you think. Most people get this one wrong. Read this thread if you want to be one of the few that gets it right.
1
As I'm more involved in organizational leadership my work is more and more "key presentations to higher level executives". Helpful content.
McKinsey makes $500,000+ on a single presentation. You can learn their simple framework below (for free):
3