Kamil Nicieja
@kamil.fyi
Ruby programmer, product builder, and author. I’m currently a lead product engineer at Plane, a Y Combinator startup making payroll tools. Read more @ https://kamil.fyi
PSA: SF Ruby is hosting an AI hackathon… and I’ll be there too! Rails has always been the go-to framework for solo developers. So here’s the big question: *exactly* how much further can we take Ruby specifically with LLMs? Let’s find out!
June 3, 2025 at 3:55 AM
PSA: SF Ruby is hosting an AI hackathon… and I’ll be there too! Rails has always been the go-to framework for solo developers. So here’s the big question: *exactly* how much further can we take Ruby specifically with LLMs? Let’s find out!
claude code is very very bad at writing tests and will always try to make the job easier for itself by cheating, BUT it’s very very good when it can iterate against a good set of tests, so perhaps this means tdd is the best way to program with an assistant
March 13, 2025 at 4:45 PM
claude code is very very bad at writing tests and will always try to make the job easier for itself by cheating, BUT it’s very very good when it can iterate against a good set of tests, so perhaps this means tdd is the best way to program with an assistant
How would you go about optimizing an existing large codebase to incorporate generative AI?
December 25, 2024 at 4:15 PM
How would you go about optimizing an existing large codebase to incorporate generative AI?
Sure, there are a lot of non-programmers sharing opinions about the future of programming right now, but to be fair, programmers have had plenty of big opinions about the future of non-programming jobs these past few years (?) too. How the turntables.
December 22, 2024 at 9:20 PM
Sure, there are a lot of non-programmers sharing opinions about the future of programming right now, but to be fair, programmers have had plenty of big opinions about the future of non-programming jobs these past few years (?) too. How the turntables.
OpenAI never lost the Mandate of Heaven.
December 20, 2024 at 6:43 PM
OpenAI never lost the Mandate of Heaven.
A tsunami warning and a tornado warning in just 10 days. Is that something that often frequently in SF? Asking, well, for myself.
December 14, 2024 at 2:54 PM
A tsunami warning and a tornado warning in just 10 days. Is that something that often frequently in SF? Asking, well, for myself.
Leaning towards “good” more often than not now—especially o1 pro.
o1 vibes based on my feeds: sometimes maybe good, sometimes maybe shit
December 14, 2024 at 5:23 AM
Leaning towards “good” more often than not now—especially o1 pro.
Everybody who complains about Ciri being the main protagonist in The Witcher 4 just lost all their credibility in complaining about games being “woke.”
December 14, 2024 at 4:51 AM
Everybody who complains about Ciri being the main protagonist in The Witcher 4 just lost all their credibility in complaining about games being “woke.”
Prediction: The more OpenAI wants to get out of their deal with Microsoft, the more we’ll see an inflation of the term AGI being thrown around.
December 12, 2024 at 10:05 AM
Prediction: The more OpenAI wants to get out of their deal with Microsoft, the more we’ll see an inflation of the term AGI being thrown around.
o1 vibes based on my feeds: sometimes maybe good, sometimes maybe shit
December 6, 2024 at 10:05 AM
o1 vibes based on my feeds: sometimes maybe good, sometimes maybe shit
OK, I didn’t expect a tsunami warning today.
December 5, 2024 at 7:06 PM
OK, I didn’t expect a tsunami warning today.
If you caught my presentation yesterday at SF Ruby & AI, here’s a quick summary. I talked about Lammy, my LLM library for Ruby that’s simple to use and treats prompts as programs rather than plain strings.
Introducing Lammy
An LLM library for Ruby
kamil.fyi
December 5, 2024 at 3:26 AM
If you caught my presentation yesterday at SF Ruby & AI, here’s a quick summary. I talked about Lammy, my LLM library for Ruby that’s simple to use and treats prompts as programs rather than plain strings.
Another SF Ruby starting soon!
December 4, 2024 at 1:39 AM
Another SF Ruby starting soon!
Just wrapped up preparing my talk for next week’s SF Ruby & AI event. Now I can look forward to it with peace of mind!
December 1, 2024 at 1:39 AM
Just wrapped up preparing my talk for next week’s SF Ruby & AI event. Now I can look forward to it with peace of mind!
It’s not a primary one but definitely one of the main three—alongside Twitter and newsletters.
Repost if BlueSky is now your primary social media site.
November 30, 2024 at 2:25 AM
It’s not a primary one but definitely one of the main three—alongside Twitter and newsletters.
This 👇
Open protocols are at greater risk for spam.
Open protocols are at greater risk for spam.
Did you know that 99% of email today is spam? Your inbox isn’t 99% spam because AI is used to filter it.
The same 99% will happen here too, but if AI researchers continue to get perma-banned for making available the datasets needed to filter it, it’s going to make this platform unusable.
The same 99% will happen here too, but if AI researchers continue to get perma-banned for making available the datasets needed to filter it, it’s going to make this platform unusable.
November 28, 2024 at 6:16 PM
This 👇
Open protocols are at greater risk for spam.
Open protocols are at greater risk for spam.
I remember how inspired I felt when Paul Graham said, “MUSTAAAAAAAAAARRRRRRDDDDDD”
November 28, 2024 at 8:08 AM
I remember how inspired I felt when Paul Graham said, “MUSTAAAAAAAAAARRRRRRDDDDDD”
I’m fine with my posts being scraped.
November 27, 2024 at 12:53 AM
I’m fine with my posts being scraped.
Interesting. I didn’t expect that they would need to be trained specifically for each type of reasoning to perform it.
A blindspot for AI reasoning engines like o1 is that they all appear to be trained on very traditional deductive problem solving for chain of thought
What would a model trained on induction or abduction do? What about one trained on free association? Expert heuristics? Randomized exquisite corpse?
What would a model trained on induction or abduction do? What about one trained on free association? Expert heuristics? Randomized exquisite corpse?
November 26, 2024 at 2:44 AM
Interesting. I didn’t expect that they would need to be trained specifically for each type of reasoning to perform it.
Wow, this is a pretty impressive idea.
A screenshot of @emilyliu.me's blogpost comments section
November 25, 2024 at 4:42 AM
Wow, this is a pretty impressive idea.
Just came across DeepSeek—it really showcases the thought process of a reasoning AI, something OpenAI's o1 doesn't quite do. This got me thinking about test-time compute this week.
November 24, 2024 at 5:09 PM
Just came across DeepSeek—it really showcases the thought process of a reasoning AI, something OpenAI's o1 doesn't quite do. This got me thinking about test-time compute this week.
No one warned me that just a few days of rain in SF could lead to flooding lol
November 22, 2024 at 11:05 PM
No one warned me that just a few days of rain in SF could lead to flooding lol
Reposted by Kamil Nicieja
Join the next #SFRuby at @sentry.io on Dec 3 to learn about Ruby&AI from @tonsoffun.bsky.social @kamil.fyi and more. We also have a (non-strict) BYOC policy this time: bring your own 🎅🎄 cookies! Who's ready to join me and bake some?=) lu.ma/q7fwrtj2
SF Ruby presents: Ruby&AI, December 🎅 edition @ Sentry · Luma
Hey, Bay Area builders and tinkerers!
From GitHub and Stripe to Chime and Stackblitz, Ruby powers teams that go HUGE with less.
Join us as we re-kindle the SF…
lu.ma
November 22, 2024 at 4:52 AM
Join the next #SFRuby at @sentry.io on Dec 3 to learn about Ruby&AI from @tonsoffun.bsky.social @kamil.fyi and more. We also have a (non-strict) BYOC policy this time: bring your own 🎅🎄 cookies! Who's ready to join me and bake some?=) lu.ma/q7fwrtj2
Re: Google having to sell Chrome. Many have criticized this decision, but for the first time in a while, there could be genuine buyers in the market who aren’t part of Big Tech like OpenAI or Perplexity. I’m not so sure this is a bad thing.
November 22, 2024 at 3:59 AM
Re: Google having to sell Chrome. Many have criticized this decision, but for the first time in a while, there could be genuine buyers in the market who aren’t part of Big Tech like OpenAI or Perplexity. I’m not so sure this is a bad thing.
I’m watching a podcast by @pragmaticengineer.com about the software development process at Linear.
It’s great to see that we at @plane.com follow many of the same practices, like the goalie system and work trials, because Linear is such an inspiring example.
It’s great to see that we at @plane.com follow many of the same practices, like the goalie system and work trials, because Linear is such an inspiring example.
Linear: move fast with little process (with first Engineering Manager Sabin Roman)
YouTube video by The Pragmatic Engineer
youtu.be
November 22, 2024 at 2:08 AM
I’m watching a podcast by @pragmaticengineer.com about the software development process at Linear.
It’s great to see that we at @plane.com follow many of the same practices, like the goalie system and work trials, because Linear is such an inspiring example.
It’s great to see that we at @plane.com follow many of the same practices, like the goalie system and work trials, because Linear is such an inspiring example.