atharva
LightNews LightNews
banner
atharvaraykar.com
atharva
@atharvaraykar.com
89 followers 150 following 50 posts
i write at atharvaraykar.com
i work @nilenso.com

yes-anding the world.
Posts Replies Media Videos
atharvaraykar.com
I wrote a post looking into multiple SWE/coding benchmarks. Many of them measure something narrower than what their names suggests.

blog.nilenso.com/blog/2025/09...
SWE-bench Verified and SWE-bench Pro What it measures How well a coding agent can submit a patch for a real-world GitHub issue that passes the unit tests for that issue. The specifics There are many variants: Full, Verified, Lite, Bash-only, Multimodal. Most labs in their chart report on SWE-bench Verified, which is a cleaned and human-reviewed subset. Notes and quirks of SWE-bench Verified: It has 500 problems, all in Python. Over 40% are issues from the Django source repository; the rest are libraries. Web applications are entirely missing. The repositories that the agents have to operate are real, hefty open source projects. Solutions to these issues are small—think surgical edits or small function additions. The mean lines of code per solution are 11, and median lines of code are 4. Amazon found that over 77.6% of the solutions touch only one function. All the issues are from 2023 and earlier. This data was almost certainly in the training sets. Thus it’s hard to tell how much of the improvements are due to memorisation.
September 29, 2025 at 7:17 AM
atharvaraykar.com
Wrote about units of work being a useful lever for getting good results from AI-assisted coding.

blog.nilenso.com/blog/2025/09...
September 19, 2025 at 9:21 AM
atharvaraykar.com
I've been poking Srihari, our most experienced engineer @nilenso.com to share his hard-earned knowledge for the benefit of others.

Even if you're not an engineering leader like me, this checklist gives a lot of insight into what makes a great engineering org.

blog.nilenso.com/blog/2025/09...
My Quarterly System Health Check-in It is essential to periodically take a few steps back from the day to day and reflect on where we are against our strategic goals. If you’re an engineering leader, a head of engineering, a director, or a VP, you likely have a recurring meeting to this effect. In this post, I propose a structure for this operational exercise (complementing a business review) that lasts 2-4 hours, every month or quarter. I see quality as solving for the Pareto front with the tangible dimensions of reliability, performance, cost, delivery and security, and the more intangible dimensions of simplicity and social structures. For each dimension, go through the list of questions below and try to answer them together.
September 12, 2025 at 6:32 AM
atharvaraykar.com
This is a good example of the weirdness of AI.
It’s a recursive paradigm shifting paradigm shift Code is data is code. Lispers get this, its turtles all the way down. Let’s walk through it. 1. Software can use AI. We write pieces of text in between code that uses AI’s thinking or knowledge access capability. Like with autocompletion, or chatbots. 2. AI can use software. It can access the filesystem and run programs on your OS, if you let it. It can access the web through search engines, or web browsers, if you let it. 3. AI can build software. It writes and executes small python scripts to analyse data when thinking. The agency and autonomy needed to build full fledged meaningful software isn’t there with AI yet, but it’s advancing quickly. AI assisted coding is pretty big, you know this. 4. AI IS software. AI is a trained neural network, and with sufficiently advanced capabilities, it can build itself.
August 14, 2025 at 7:15 AM
atharvaraykar.com
Why Does AI Feel So Different?

An enjoyable read from my colleague, Srihari.

We've been talking about why this disruption feels different from other recent technological disruptions and he captured a lot of that really well in this post.

Link: blog.nilenso.com/blog/2025/08...
Image description While Kuhn doesn’t go into it, the technological diffusion, and the economic impact of scientific revolutions are better studied through the GPT (general purpose technology) paper from Bresnahan & Trajtenberg in 1995. > “General Purpose Technologies (GPTs) are technologies that can affect an entire economy (usually at a national or global level). They have the potential for pervasive use in a wide range of sectors and, as they improve, they contribute to overall productivity growth.” And Calvino et al in June 2025, finds that AI meets the key criteria of a General Purpose Technology. It’s pervasive, rapidly improving, enables new products, services and research methodologies, and enhances other sectors’ R&D and productivity.
August 14, 2025 at 7:15 AM
atharvaraykar.com
Question for the "launch many parallel claude agents to go faster" crowd: Why isn't that the same problem as the Brooks's law all over again?
wikipedia screenshot of brooks's law page.
July 28, 2025 at 2:14 PM
atharvaraykar.com
Several people ask me about how I'm keeping up with all the AI things and finding signal in this noisy landscape. I wrote a guide explaining this.

blog.nilenso.com/blog/2025/06...
Table of Contents General guidelines Starting Points Official announcements, blogs and papers from those building AI High signal people to follow News and Media Esoterica Do I chug water from a firehose?
July 4, 2025 at 9:00 AM
atharvaraykar.com
I was seeing @karpathy.bsky.social's Software 3.0 talk, as one should.

Surreal to see him recommend my writing to make a point on AI-assisted coding

"Is that me on TV??" moment
Andrej Karpathy on the YCombinator startup school stage talking about my AI-assisted coding blog post
June 19, 2025 at 1:17 PM
atharvaraykar.com
Buddhist Teacher Geshe Langri Tangpa (circa 1100 AD)
🤝
XXXTentacion (2018)
In brief, directly or indirectly, I will offer help and happiness to all my mothers, And secretly take upon myself All their hurt and suffering. We ain't playin' with y'all niggas, man, you heard? Sometimes you just gotta catch chlamydia on these niggas, G shit You know what I'm saying? Gonorrhea, all of that shit All of that shit I catch all diseases in the world So the world don't have no more diseases, you feel me? G-shit Yeah, yeah (P.Soul on the track)
January 18, 2025 at 9:01 AM
atharvaraykar.com
TIL about the Modi (मोडी / 𑘦𑘻𑘚𑘲‎) script, used for Marathi up until the 1940s.
In 2025, it looks like AI generated Devanagari.
January 4, 2025 at 1:41 PM
atharvaraykar.com
i quickly concocted a writer's block unblocker
(with @tldraw.com computer)

it takes an oblique strategy (from the brian eno et al card deck) and uses it to provide unhinged critique of the essay you are working on to help you break out of a rut

link to program: computer.tldraw.com/t/4KoB33nFEr...
December 29, 2024 at 5:23 PM
atharvaraykar.com
menu highlights
google's menu highlights for a thindi joint. the first highlight is half eaten masala dosa accompanied by a picture.
December 1, 2024 at 6:49 AM
atharvaraykar.com
art in a white wall big box chamber is gazed upon for long, an object of loving attention

art outside the door of hae kum gang korean restaurant gents toilet is ignored among the din of swinging doors and flushing toilets
dreamy surreal distortion of a bicycle framed outside hae kum gang korean restaurant gents toilet
November 30, 2024 at 3:32 AM
atharvaraykar.com
chai darkening ideas
- add fine mud
- spray spray tan
- mix in water from a lake in peenya
- chocolate
- blend with chyawanprash
November 27, 2024 at 4:10 PM