Yet another week of the fellowship comes to an end. This week was special, not only because of the festivities around Holi(day) but also because we're officially into the Build Phase now 💪
Well, technically, the DeFi Presentation that I wrote about in the last week's essay happened this week, but I couldn't hold the excitement back then, so here's a link back to the previous essay :)
We also got to listen to a panel on DeFi and Cross-chain Applications with some of the most brilliant minds of DeFi, namely Samyak Jain (InstaDapp), Francesco Renzi (Superfluid), Vaibhav Chellani (Socket), and Arjun Bhuptani (Connext). All of them were extremely passionate and proud of the work that they did and it reflected in the way they spoke about their projects. A lot of what they spoke about was very new to me and I was humbled by the sheer depth with which they talked.
One particular project that I still remember from the discussion though is InstDapp's Avocado Wallet. Avocado is a smart-contract wallet, which means that it abstracts away the Externally Owned Wallets (EOAs) which are controlled by a private key to a smart contract that takes care of it all (see ERC-4337).
Another discussion that I remember, particularly because of how good the advice is, is the need for ideating and building projects to fix things that one personally feels about, rather than chasing 'the current thing'. Couldn't agree more! Speaking of ideas and projects, I think I should also talk about what I'd be building next, and some of the other ideas that didn't make through right now.
At the start, I was all hyped up about what I could possibly work on. I recently read about Verkle Trees and they looked tempting. Why not use them everywhere? So I planned to take on the challenge of implementing these over Merkle Trees in some of the places where it is currently being used.
Another idea that was lingering around at the same time was using ZK proofs to keep a decentralized record of all the inference processing that a Machine Learning model does. This would secure the data/model by keeping them private and by providing ZK proofs, we can show that the data/model is actually what I claim it to be.
I mean I have 3-4 weeks. How hard can any of these ideas get? Turns out, pretty hard. Both of them. And when I say 'hard', I mean it both in terms of the complexity of the solutions and the time taken to implement them even if the solution was right in front of me. So, definitely not something to pick up hoping to build something viable in the limited time of the next few weeks. So I started to look for another idea and this now sounds to me like…
I saw the below Notion Doc below posted somewhere and looked through it. Something along the lines of "allow human creators to cryptographically sign every piece of content they create using an Ethereum key pair that they control” caught my eye.
I've heard the likes of ChatGPT watermark text created by them to algorithmically distinguish them from a human-generated piece of writing; And I thought to myself, “What if we can do something similar for images? Leave some kind of cryptographic watermark on your images to distinguish that you are the original creator of this image?" Sounds dope to me. And I asked myself:
Is the solution to this problem complex to find out? - Nope. Already have most things figured out (Shoutout to QEDK for helping with this)
Is it something for which I can build a working app in the next few weeks? - Definitely.
All points checked. So this would be it! I'll be building this for my EIF project. BTW, I know I did not do a good job explaining the solution and how I am thinking to go about with it and that is deliberate (you see, a feature, not a bug) so that the reader (yes, you) comes back for the next post next week and see how I am going about this idea.
So, excited about the next week's update? Meeee toooooo! LFG 🤘
Until next time,