Hello Reader,

If you’ve been following my updates, you know I’m deep in the world’s largest hackathon right now.

But this isn’t just a coding challenge for me; I’m treating it like a real startup sprint.

In this week's newsletter, I share everything:
– What we’re building now
– How we’re building it (tech stack, tools, process)
– And most importantly, why we’re making the decisions we are

But first, some backstory.

We’ve already killed four ideas.

Yes, really.

Since Virgil and I became cofounders, we’ve explored, prototyped, and launched four different products and shut each one down.

The last idea we were working on was Shopify SEO Autopilot.

We spent a full week researching it, doing competitor analysis, breaking down features, reading their reviews, and reaching out to Shopify store owners.

The results?

Too crowded. Too noisy. We’d be entering late and without a strong edge.

The pivot: A video ad platform

Around the same time, my co-founder, Virgil, connected with Victor, an experienced ad agency founder who has been running his agency since 2019.

We saw an opportunity to turn his ad creation process into a product.

We had several meetings to validate the idea. We also did initial research and competitor analysis.

The biggest advantage of this idea is the deep domain knowledge that Victor has.

On June 8, we decided to pivot to this new idea of building a video ad generation platform.

We spent the next few days working on various parts of the application. Virgil focused on user flow and wireframes, Victor concentrated on prompt engineering, and I worked on tech.

4 POCs to find the best tech stack

I spent last week exploring and building 4 proof-of-concept apps:

1. Next.js full-stack - frontend and backend in a single application
2. React + Node.js (Express) - frontend and backend separated using the same underlying tech
3. React + FastAPI - Python Fast API with React frontend
4. React + Django - A popular Python framework, but too much overhead for our use case

After testing each one, we decided on React (frontend) + Python + FastAPI (backend).

Python integrates more effectively with AI and offers faster performance. Node.js and Express have the advantage of maintaining the same technology across the frontend and backend, resulting in a lower learning curve for new team members.

However, after comparing all options, we decided to use Python with FastAPI for the backend.

For the web servers, I conducted similar research, comparisons, and proof-of-concepts before deciding to go with AWS Lightsail. Input from my friend Piyush Sachdeva was very helpful in this decision.

And my friend Vishal suggested applying for the $1,000 AWS Startup Credit, which I did and received.

The tools behind the scenes

Here’s the exact toolchain I’m using to build this, fast:

• Three AI coding tools: Bolt .new, VS Code + GitHub CoPilot and Cursor editing the same project.
• Figma for wireframes, Figjam for brainstorming
• OpenAI, Anthropic, Elevenlabs, Flux1, Kling, Veo 3 etc
• Loom, Google Docs, Google Chat for collaboration

Sharing it all in public

I’m documenting the entire hackathon journey with:

• 9+ videos so far on YouTube
• Frequent updates on X and LinkedIn
• Real behind-the-scenes with pivots, struggles, and wins

I want to share with other builders, especially those in this hackathon, the messy, non-linear, learning-heavy kind of building that we don’t talk about enough.

You can find all my videos here.

​

What’s next

We’re about halfway through the hackathon. In the next 15 days, we’ll:

• Finish the full v1
• Launch with a small closed beta
• Gather feedback from users and refine fast

We already have interest from a few business owners and agency leads, and we’re excited to get it in their hands.

If you’re in the hackathon too, I’d love to hear what you’re up to.

And if anything I’ve shared is useful, feel free to reach out. I’m happy to help.

Talk soon,
Vinod