Efe Karakus
@efekarakus.com
I like sharing what I learn about building better software.
Engineer @ AWS
Engineer @ AWS
Reposted by Efe Karakus
It bothers me to no extent that the media plays how Klarna whistles
Klarna wants to the headlines to say "AI-enabled Klarna"
It would have halved its workforce without AI because it overhired 2x in 2020-21 and never corrected till now
But this is how they get front pages
Klarna wants to the headlines to say "AI-enabled Klarna"
It would have halved its workforce without AI because it overhired 2x in 2020-21 and never corrected till now
But this is how they get front pages
October 13, 2025 at 5:19 AM
It bothers me to no extent that the media plays how Klarna whistles
Klarna wants to the headlines to say "AI-enabled Klarna"
It would have halved its workforce without AI because it overhired 2x in 2020-21 and never corrected till now
But this is how they get front pages
Klarna wants to the headlines to say "AI-enabled Klarna"
It would have halved its workforce without AI because it overhired 2x in 2020-21 and never corrected till now
But this is how they get front pages
Is 10pm on a Sunday a good time to post about your new tech blog? www.efekarakus.com/2025/09/21/l...
Here are some basic thoughts if you are on a team vending functionality for AI agent builders
Here are some basic thoughts if you are on a team vending functionality for AI agent builders
Leave integration choices to agent builders
If you’re building an agentic product, especially in a large tech company, you’ll likely need to integrate with multiple partner teams that offer unique context or actions that enhance your agent’s pe...
www.efekarakus.com
September 22, 2025 at 5:12 AM
Is 10pm on a Sunday a good time to post about your new tech blog? www.efekarakus.com/2025/09/21/l...
Here are some basic thoughts if you are on a team vending functionality for AI agent builders
Here are some basic thoughts if you are on a team vending functionality for AI agent builders
Reposted by Efe Karakus
Confounding that we have the simultaneous narratives of "AI will replace your jobs for 1/10th the price" in parallel with "there just are not enough people to build fast enough, we need you to work 80% harder."
Also, it was exactly the same 15 years ago except replace AI with "outsourcing."
Also, it was exactly the same 15 years ago except replace AI with "outsourcing."
My friends. This is not a good idea: www.wired.com/story/silico... - take it from every single wisdom tradition in the history of the world. You can love your work, but your work cannot love you back. Have more respect for yourself and your peers. Work hard. Do good work.
Silicon Valley AI Startups Are Embracing China’s Controversial ‘996’ Work Schedule
In an industry once known for cushy perks, some founders are now asking staff to commit to a 72-hour weekly schedule. You’re either in or you’re out.
www.wired.com
July 23, 2025 at 4:50 PM
Confounding that we have the simultaneous narratives of "AI will replace your jobs for 1/10th the price" in parallel with "there just are not enough people to build fast enough, we need you to work 80% harder."
Also, it was exactly the same 15 years ago except replace AI with "outsourcing."
Also, it was exactly the same 15 years ago except replace AI with "outsourcing."
Reposted by Efe Karakus
✏️ Wrote a blog post on what small teams building LLM-based applications should actually focus on each sprint: efekarakus.com/2025/07/06/e...
Experimentation over capabilities
In traditional software engineering, we create customer value by working on small, focused features delivered frequently. When building products where generative AI is a key component, I’ve noticed or...
efekarakus.com
July 7, 2025 at 3:45 AM
✏️ Wrote a blog post on what small teams building LLM-based applications should actually focus on each sprint: efekarakus.com/2025/07/06/e...
✏️ Wrote a blog post on what small teams building LLM-based applications should actually focus on each sprint: efekarakus.com/2025/07/06/e...
Experimentation over capabilities
In traditional software engineering, we create customer value by working on small, focused features delivered frequently. When building products where generative AI is a key component, I’ve noticed or...
efekarakus.com
July 7, 2025 at 3:45 AM
✏️ Wrote a blog post on what small teams building LLM-based applications should actually focus on each sprint: efekarakus.com/2025/07/06/e...
Reposted by Efe Karakus
Great talk about avoiding rewrite trap by @skamille.themanagerswrath.com at @leaddev.com #LDX3, and my favourite: "The Computers are not your stakeholders" 👏
leaddev.com/software-qua...
leaddev.com/software-qua...
Escaping the rewrite trap
Gain practical frameworks to assess rewrites vs. evolution, plan successful migrations, and align teams and stakeholders without losing momentum.
leaddev.com
June 30, 2025 at 4:11 PM
Great talk about avoiding rewrite trap by @skamille.themanagerswrath.com at @leaddev.com #LDX3, and my favourite: "The Computers are not your stakeholders" 👏
leaddev.com/software-qua...
leaddev.com/software-qua...
Reposted by Efe Karakus
New blog post, with some career advice: brooker.co.za/blog/2025/06...
Career advice, or something like it - Marc's Blog
brooker.co.za
June 20, 2025 at 6:14 PM
New blog post, with some career advice: brooker.co.za/blog/2025/06...
I’m really enjoying sketch.dev’s blog posts on agentic AI development https://sketch.dev/blog/push-pull-respond-restart
sketch blog: Push, Pull, Respond, Restart
sketch.dev
June 7, 2025 at 11:23 PM
I’m really enjoying sketch.dev’s blog posts on agentic AI development https://sketch.dev/blog/push-pull-respond-restart
i wanna be grug grugbrain.dev
The Grug Brained Developer
grugbrain.dev
April 28, 2025 at 7:02 PM
i wanna be grug grugbrain.dev
Reposted by Efe Karakus
Product evals are misunderstood. Many teams think that adding another tool, metric, or llm-as-judge will solve all their problems and save their product. But that just dodges the hard truth and avoids the real work. Here's how to fix your process instead.
eugeneyan.com/writing/eval...
eugeneyan.com/writing/eval...
An LLM‑as‑Judge Won't Save Your Product—Fixing Your Process Will
Applying the scientific method, building via eval-driven development, and monitoring AI output.
eugeneyan.com
April 23, 2025 at 2:45 AM
Product evals are misunderstood. Many teams think that adding another tool, metric, or llm-as-judge will solve all their problems and save their product. But that just dodges the hard truth and avoids the real work. Here's how to fix your process instead.
eugeneyan.com/writing/eval...
eugeneyan.com/writing/eval...
New blog post with thoughts on the promotion process and growth for senior engineers: efekarakus.com/2025/04/12/o...
On promotion and growth beyond senior
As I progressed from a junior to senior software engineer within Amazon, I came to view leveling up as a function of wisdom - f(wisdom) - where as your applied or practical knowledge increased you wer...
efekarakus.com
April 12, 2025 at 2:54 PM
New blog post with thoughts on the promotion process and growth for senior engineers: efekarakus.com/2025/04/12/o...
Unreasonable Hospitality by Will Guidara is one of the most inspiring management books I read recently 😍. It helped me pin down the perfect term: “being unreasonable” for what I enjoy doing most at work. a.co/d/d0NH6hb
Unreasonable Hospitality: The Remarkable Power of Giving People More Than They Expect
Unreasonable Hospitality: The Remarkable Power of Giving People More Than They Expect [Guidara, Will] on Amazon.com. *FREE* shipping on qualifying offers. Unreasonable Hospitality: The Remarkable Power of Giving People More Than They Expect
a.co
April 6, 2025 at 10:00 AM
Unreasonable Hospitality by Will Guidara is one of the most inspiring management books I read recently 😍. It helped me pin down the perfect term: “being unreasonable” for what I enjoy doing most at work. a.co/d/d0NH6hb
Wrote a short blog post on how goodput can degrade ↘️ as offered load to our system increases ↗️ www.efekarakus.com/2025/01/08/g...
Goodput degradation in rate-limited systems
In this post, we’ll explore how “goodput” (the rate of useful, successful work completed) degrades under increasing load, particularly in multi-step workflows that depend on rate-limited services.
www.efekarakus.com
January 9, 2025 at 2:40 AM
Wrote a short blog post on how goodput can degrade ↘️ as offered load to our system increases ↗️ www.efekarakus.com/2025/01/08/g...
I really enjoyed Poor Charlie’s Almanack (a.co/d/0DBK4qg). My favorite talk, which summarizes most of the content in the book, is also available on YouTube: youtu.be/lqqRzNC3QyU
November 28, 2024 at 12:01 AM
I really enjoyed Poor Charlie’s Almanack (a.co/d/0DBK4qg). My favorite talk, which summarizes most of the content in the book, is also available on YouTube: youtu.be/lqqRzNC3QyU