So much work!

Oh man, where do I even begin? Suddenly there has been so much work that has come up. This is partly because of the part-time job, ETHDenver coming up, and also a couple of hackathons beginning simultaneously. Perhaps another reason is the scope of ambition increased because of Claude Code. And maybe also because while I might not have realized it earlier, I had been slacking / procrastinating a bunch in the past. For instance my YouTube time tracker has gone down from about 4-5 hours a day to less than 1 hours a day! Similar dopamine resets for mindlessly scrolling Instagram reels as well. All good news, it would seem..

Anywho the rest of this post is sectioned into major themes like in the previous post;

Part Time Role

So, this really picked up. Like I mentioned last time, the plan was to use Claude Code to finish up all my pending tasks. Turns out there was A LOT pending. Even with Claude Code I spent 20+ hours (my contract is only part-time) and I probably need another week of 20-ish hours. This just means that if I were to do this manually, this was like a 2-month operation! This kind of ties into something I really need to improve on: estimation. I often end up overestimating my capabilities and motivation, and underestimating the actual work required, both combined they really fuck things up. So much so that in my role, I actually ended up on really thin ice but that is probably a story for when I am off that ice.

That being said the current tasks there are to refactor my ad-hoc Flask application into a proper, structured FastAPI services module, along with other modules for crew AI and also using a message queue for handling large jobs asynchronously and to split them up. Other than that I also had to refactor my Next.js application into a React + Vite application. Again I probably know like 30% of all these technologies, so the two-month estimation I gave above should probably be increased by another month if I were to do it all by myself and make all the decisions myself. However with Claude Code, not only does it actually write all the code and I'm just supervising / reviewing PRs, But I can also chat with it and determine the best way forward, plus fill in gaps in my own knowledge. Truly a game changer!

Denver

The other big thing taking up mindshare this week was the trip to Denver. ETHDenver is next week and I plan on driving to Denver, though the functionality of my car is somewhat up in the air. Fortunately because of my flatmate, I was able to give my car to a community college where they are working on it as part of their auto / mechanics course; no labor charges! Plus they also get parts at a steep discount on account of being a college. I did receive my car on Thursday and I plan to make the drive from Saturday onwards. The car also needs new tires. While I want fancy tires, I think I'm just gonna get the cheapest ones from Walmart.

But why Denver? Well for ETHDenver of course! There is actually a lot going on during this ETHDenver for me personally. Of course there's the hackathon where I am planning on building Unhinged. But other than that this is also the first in-person crypto meet where I can start shopping my idea of an AI Film Festival (more soon!), which I want to host during Devcon in Mumbai; I am from Mumbai! Obviously I'm building my own platform for managing the event and that platform will have multiple features. One of those features can potentially be battle-tested at ETHDenver as a proof of concept. Will have more to say about this in the next week's post. Other than that, it's a chance to meet all my online friends in person again! Since crypto has been really down so far, I don't know if many people are actually going to show up in India in November. So frankly this might actually be the only chance I get to meet everybody in person this year.

Eh, who am I kidding? Of course people are going to show up at the other conferences as well!

khaaliDimaag

The other thing that has had a serious improvement this week has been khaaliDimaag. And this is all thanks to Trxnt with his baseline system, what an unlock! To be honest, because of Claude Code and the baseline system, I think I might be able to create the company entirely solo and not need to hire anybody for a very long time. Want to? Sure. Need to? Probably not. Key distinction there.

So what was achieved this week for the company? One, the baseline system allows for is taking care of all the non-technical parts of a project. And honestly, this was like a huge blocker for me; because oftentimes I would meet folks and require 30-60 mins of their time to actually explain the concept (which has been down from 150-180 mins) but still not ideal. However, with that system a lot of my ideas and thoughts were concretely articulated and documented, not perfect, but about 80% of the way there, and that's a great push forward! Now the plan is to document all my varied projects, break them down into specific roadmaps and feature sets, and then get buidling!

Two, I also had been thinking about a governance system that is public and coordinated via github discussions. The basic idea is that there are 16 categories an idea or a feature goes through all the way from thought-bubble to release-announcement, while considering implementation details, marketing and onboarding plans, security concerns and mechanism design, and any conflicts or domain expansion possibilites! It still is in a PR stage (so probably link to it in a couple of weeks), but the core categories, their descriptions, and the flow through the categories are pretty much set now. So, progress!

And then three, one of the issues also was having a manipulatable logo on hand. The current logo was drawn out (on one inspired night) by pencil, then a photograph taken of it, and finally drawn over within preview using a mouse, and finally exported. Now I know there's better ways to move drawings to digital, but when I first made it about four or five years ago, I didn't know better and now I can't find the original piece of paper or the photo. That being said, Figma. Oh my god. I was using masks on the logo to get different colors (the original image was #000), but Figma vectorized the entire thing and now I have a proper SVG!! I do need to simplify certain paths but, I actually have something to work with now; and I can also do the line animations I have been thinking about to represent the different projects under khaaliDimaag. Exciting times ahead!

Figma

Speaking of Figma, I have been trying to get better at Figma, partly because its a good way to find expression for my ideas and partly because I kinda want to be a one man army; and this week I realized that I have actually improved a lot! I know a bunch of best practices now, and also where everything is on the interface (plus hotkeys!), and I can also build stuff out quickly as well. Within limits of course, but much better than say a year ago when I seriously started trying. The realization came when I was on a call with my friend for the khaali Film Festival (more next week!) and built out the UI of the landing page on the call within ~30 mins. Pretty cool!

Hackathons

Let's keep the final section of this post for the hackathons, since there are a bunch all of a sudden.

Hack Money 2026

First, the one that just ended; well didn't really win this one, but the revealation here was claude-code! I actually did ship the project in about 3 days (albeit 14-16 hour days) and while slightly buggy, the code is pretty solid, and I barely looked at it! There is definitely a need for better documentation though, but like I mentioned above, the baseline system can fill in that gap.

ETHDenver

This is the current meta now, though I am not sure if or how much I can actually do here. Since this involves going to Denver, trying to pitch the khaali Film Festival (again, more on this next week), shipping a few features to test live for it, and also general networking, connecting, and meeting online friends, I wonder how much time I will actually get to build the project for the buidlathon.

Having said that, I am planning to build out Unhinged here. Unhinged is largely an exploration in prediction markets and maybe also eliminating a need for the UI. What I am trying to get to is an attempt to answer the question "What does a particular group of people, think of a particular opinion, at a given time?" by leaning into the wisdom of the crowds aspect of prediction markets. A lot of the ideas are inspired by the a16z crypto episode on prediction markets. But, for more information, probably check out this thread on X?

PL_Genesis

Finally, the longest of the current hackathons, and perhaps the most ambitious in eventual scope; also a key piece of the khaaliDimaag ecosystem. This is a Protocol Labs hackathon, so the focus here is to knock off a rather serious project - dApp Ideas. At its core, dApp Ideas is a tool to manage personal experiments, but tying it to web3, this becomes a real powerful tool to help DeSci by helping conduct global scale studies. I shall have more to say about this in the coming weeks as I am still figuring out the final nuances of the project, and also because I have been yapping for too long already.