Siddhartha Golu

@siddharthagolu.com

75 Followers 51 Following

I read books to understand myself. Infrastructure nerd. Co-Founder & CTO, one800.help. Writes sporadically at siddharthagolu.com

HeliBoard app screen for contributing gesture data
Matt Levine's column's starting paragraph: "PR insider trading Every financial services firm will tell its new employees, in their first week or two of work, the same thing: "Be careful with our confidential information. Don't tell anyone outside of the firm what you're working on. Don't talk carelessly in elevators. Don't leave your laptop unlocked. Don't even tell your spouse what you're working on." Every new employee will take that to heart, for a week or two; they will go home and their boyfriends or girlfriends will ask "so what did you do at work today" and they will say "I can't tell you it's a seeeeeeeeecret" and their boyfriends and girlfriends will be like "aww that's cute." And then they will work there for a few months, and some of this scrupulousness will get sanded away, and they will come home after a long day at work and be like "ugh John Smith at Amalgamated Sprockets is really getting on my nerves, he keeps demanding revisions on the model for the merger with Consolidated Widgets that they want to announce next Wednesday," and their boyfriends or girlfriends will be like "yeah you complained about that yesterday too." "
A screenshot of my discover weekly playlist where I had to hide all the recommendations because they were so bad.
A comment on hacker new which says: I think, is that everyone who needs small software to manage a todo list or synchronize files, or whatever "normal" people do, will end up with bespoke personalized software written by their own AI People can't be bothered to cook for themselves, and often order crappy, unhealthy food that costs 10 times as much just so they don't have to cook. Now they're going to build their own software every time they think they need an app..?
Bluesky settings for turning off auto play videos
A reddit comment on the thread "why do people consider Wes Anderson as pretentious" which says: "The problem with Wes Anderson isn't that he has a very specific style (you wouldn't go to a Monet retrospective and complain about how many times he painted water lilies), it's that we don't have a hundred Wes Andersons all pursuing their own styles. As it is, it's like crawling through a desert and finding a fountain that only dispenses your least favorite soda. The problem isn't the soda, the problem is the lack of choice and supply."
A screenshot of a work-in-progress web UI to generate speech using mlx-audio server
A comment on HN (Part I) in response to Mitchell Hashimoto's post on his AI adoption journey. It says: "LLMs are not for me. My position is that the advantage we humans have over the rest of the natural world, is our minds. Our ability to think, create and express ideas is what separates us from the rest of the animal kingdom. Once we give that over to "thinking" machines, we weaken ourselves, both individually and as a species. That said, I've given it a go. I used zed, which I think is a pretty great tool. I bought a pro subscription and used the built in agent with Claude Sonnet 4.x and Opus. I'm a Rails developer in my day job, and, like MitchellH and many others, found out fairly quickly that tasks for the LLM need to be quite specific and discrete. The agent is great a renames and minor refactors, but my preferred use of the agent was to get it to write RSpec tests once I'd written something like a controller or service object. And generally, the LLM agent does a pretty great job of this. But here's the rub: I found that I was losing the ability to write rspec. I went to do it manually and found myself trying to remember API calls and approaches required to write some specs. The feeling of skill leaving me was quite sobering and marked my abandonment of LLMs and Zed, and my return to neovim, agent- free."
Matt Levine's new post on xAI and SpaceX merger
Three apps in the play store, all have app icons which are block letter on vibrant colorful backgrounds
A HN user posting his experience with Waymo
Bertrand Russell's quote which says: "Make your interests gradually wider and more impersonal, until bit by bit the walls of the ego recede, and your life becomes increasingly merged in the universal life. An individual human existence should be like a river - small at first, narrowly contained within its banks, and rushing passionately past rocks and over waterfalls. Gradually the river grows wider, the banks recede, the waters flow more quietly, and in the end, without any visible break, they become merged in the sea, and painlessly lose their individual being."
A Firefox user reviews my extension to hide like metrics on Twitter/Bluesky: "Works as expected for Bluesky, and you can still view reposts/likes by clicking on the corresponding buttons on individual posts, but the tedium makes it a lot less tempting. Thanks <3"
Unsolicited github spam email asking to try yet another tool
Supertonic running in the browser locally via WebGPU on Firefox
Expertise of the Igalia team: Browsers and Client-side Web Technologies, Multimedia, Accessibility Tools and Technologies, Compilers and Programming languages, Virtualization and Cloud, Embedded Linux and Device Drivers, Graphics Pipeline and Rendering Technologies, Linux Kernel.
A comment on Lobste.rs explaining the biggest problem with Slack/Zulip is that they enforce a completely new account per community.
A reddit comment which recommends that you should sit 2/3rd the way back in the theaters for optimal sound. The entire comment: "I learned this from my sound design class where my prof was an audio engineer. He had done work for all kinds of venues and had some crazy wizard level knowledge about audio. He used only two stereo speakers to make a surround sound audio experience. He directed them specifically and placed chairs in specific spots and if you sat in those two chairs the audio was literally mesmerizing. He then got on the whiteboard and used a bunch of jargon that was over my head while explaining in detail why it's better to sit 2/3rd the way back rather than perfectly halfway. Something about soundwaves and how they travel but honestly I can't remember exactly why."
An update from Play store which mentions that the corner radius for app icons on Google Play is increasing from 20% to 30% on March 31st, 2026.
Screenshot of a work-in-progress time tracker Firefox extension. Shows overall time spent at the top, followed by a list of URLs in decreasing order of time spent in each of them.
A screenshot of the OpenObserve dashboard, shows 6.9 billion events, 8.81 TB ingested size
Screenshot of AntennaPod wrapped which showcases my Top 5 podcasts from 2025. They are: 1. The Seen and the Unseen 2. Acquired 3. Lex Fridman Podcast 4. The Changelog 5. Everything is Everything
A movie poster for the film "Cactus Pears". Shows two men with their eyes closed.
A screenshot of a Firefox extension that I'd created named "Zen Twitter/Blesky". Shows 2 reviews and 31 users.
A poster for the film "Jugnuma/The Fable (2024)". Shows a happy family under an open sky.
A funding news for an astrology based startup in India, uses AI
A screenshot from a changelog of a software update which says: "New in this release: we've fixed bugs and added various other enhancements."
A screenshot of the bare-bones markdown reader app for Logseq, shows my graph's today's journal entry
FreshRSS setting to mark an article as read, includes my settings to remove GitHub's spam
Reddit's notification settings, showing you have to turn off notifications for each community
A screenshot of y-websocket library's documentation hosted on GitBook which shows a "Copy Page as markdown for LLMs" button
A paragraph from a recent piece by Matt Levine says: "I have described Altman's marketing approach as "business negging." If you say "our product is good and will make the world a lot better," that is sort of ho-hum; everyone says that, so your audience will discount it. But if you say "our product will probably make the world a lot better but could destroy humanity, tee hee aren't I a little rascal," that's exciting! People are drawn to danger, and if you go around warning people that your product might destroy humanity then that strongly suggests that you think it's powerful."
A series of posts marked as "Show less like this"
Bluesky's screenshot of a post getting disappeared from Discover when you mark "See less like this"
A screenshot from Changelog's Zulip discussion around prison reforms: "I have been thinking about this for the last couple of days and a thought occured to me. A child is born in your town in a hospital not too far away from you, maybe the same hospital you were born in. Is raised in your culture and traditions, goes to public school just like you did. Watched the same TV shows as you, listened to same songs we all do, and participated in the same culture as we all do. One day this person is arrested for whatever reason. Society doesn't look at this person and say "how have we failed this person, what did fail to provide him or her that they had to resort to crime or drugs. What kinds of messages did our culture and education system feed them which resulted in this etc". Instead society says "this person failed us, they chose to harm our society. This is an affront to our society and they must be punished harshly as possible." I think this is at the core of what's wrong with our so called justice system. Note: It's obvious the justice system in the USA is collapsing rapidly but this notion applies to most other countries as well."
A screenshot of a comment from Lobste.rs which says: "I had some plans to finish one of my projects with data pipelines (using local environment), but russia decided to shell Kyiv with ballistic missiles and drones this night, so most of the night was quite awful; dogs were shocked, humans crying, all that stuff. No direct hits (at my building), but one was quite close in the neighbourhood. So l'll put some effort in pinging my colleagues and getting some sleep afterwards. Data pipelines - optional."
A screenshot from Google AI Studio showing marathi language in logs
A screenshot of amazon now saying the order got delivered in 39 mins
A quotation from the book "A history of western philosophy" by Bertrand Russell, which reads: "Philosophy, as I shall understand the word, is something intermediate between theology and science. Like theology, it consists of speculations on matters as to which definite knowledge has, so far, been unascertainable; but like science, it appeals to human reason rather than to authority, whether that of tradition or that of revelation. All definite knowledge-so | should contend-belongs to science; all dogma as to what surpasses definite knowledge belongs to theology. But between theology and science there is a No Man's Land, exposed to attack from both sides; this No Man's Land is philosophy. Almost all the questions of most interest to speculative minds are such as science cannot answer, and the confident answers of theologians no longer seem so convincing as they did in former centuries."
Screenshot of the latest Qwen3 models comparison across benchmarks with other closed/open source models
Screenshot of a poem which reads: "I’m scared. No, not because there’s a pandemic going on, even though that itself is enough to make one lose their mind. There’s another deadly virus crawling its way into my whole being. Slowly eating away at my sanity, feasting on my thoughts, mutating inside my head as if a colony of ants was given an open invitation to a room full of sugary syrup. I’m disintegrating. Funny how the same thoughts, who once used to tell me there’s an ocean of possibilities out there; are slowly drowning me into the same pool, replacing me, devouring me bit-by-bit as if my feet are tied with stones of expectations and I keep swimming to the bottom, mistaking it for the shore. Splash. Splash. Splash. I see people making art, saving lives, reading books, writing novels, cooking food, running home marathons - resolving to turn this gloom into resilient happiness. And I can’t help but feel disgusted at my curled up crying self in the bedroom corner, struggling to tell apart midnight from noon."
A screenshot of a time tracker application showcasing the amount of time it took me to read the book "The complete story of civilization: Caesar and Christ"
a screenshot of an api response which has weird and inconsistent variable naming
A screenshot of the new landing page of Anki
A screenshot that says "You have successfully unsubscribed. You have been resubscribed successfully! As per your request, {emailID === 'you'? 'you': emailID}} {emailID === 'you'? 'have' : 'has')} been permanently removed from our mailing list As per your request, {{emailID}} will continue to receive emails from us. There has been a mistakeResubscribe Me!"
A weird captcha on Twitter where you're asked to play "find the match" but increasingly harder to do for humans as well
Screenshot of a reddit thread says: "Their servers might be collapsing but they just couldn't care less Imao They very likely made billions today on Nvidia puts. Their parent company is a hedge funds you can bet that those finance bros knew how this plays out. They likely made bigger profits today than OpenAl will do in years."
Screenshot says: Trump vs Melania: TRUMP coin plummets over 50% after MELANIA meme coin launch
The first rule of medicine is "first, do no harm". In metrics, monitoring, and instrumentation, it is the same way. Scars from production for me have taught me that: • All calls to instrumentation (logging or metrics) should be considered as screaming into the void in the worst case. • All calls to instrumentation should be non-blocking (including, potentially, being turned off entirely). • You should never, never, never rely on the logging output or format for any business process (say, via clever use of Datadog log filters triggering alerts for your CS team); if you need that functionality, treat it as a first-class business object and not a happy accident of monitoring. • You should never, never, never put logging and metrics in the same storage as your business-sensitive stuff. An over-eager logger should not be able to take space meant for paying customers. • You should pick a reasonable retention window (probably mere days) for logs/traces and (probably months) for metrics. If you cannot decide and justify a particular window, you need to learn your business better. • You should aggressively prune and ignore (for the purposes of stacks that allow indexing) logs that are not useful. It is not useful to have gigabytes of careful indexing of operation complete, especially if you also have traces. • You should keep a close eye on your developers and periodically sweep for logging that escapes into production. I have seen thousands of dollars spent logging messages that had the level set incorrectly because, basically, it looked prettier in the debug output in local dev. • You should be sweeping through your captured exceptions and fixing them, and if your org can't do this you should stop tracking them entirely. • You should be fixing exceptions as soon as they crop up, because they're probably covering up some other brain damage in your system. Failure to do this will massively slow down efforts to fix tricky customer-facing bugs, and I guarantee those bugs will …
I traveled by sleeper train in india, around 2010. They had beds, but every time I woke up, there were 3-4 Indian dudes that had come into our cabin and climbed on my bed to get some shut eye. It wasn't threatening or anything, just a wild experience and insightful lesson in cultural differences. Strange men climbing into one's bed sounds very threatening to me..? It's less invasive, but still crazier than it sounds. Indian railways changed the base sleeper cars into free-for-alls by changing sleeper cabin classifications and stopping verifying tickets. Now you have people buying the cheapest tickets (unreserved general) and swarming the "reserved" sleeper cabin berths. [0] They're just over- cramming the trains.