Under the Hood: The Technologies Powering Our AI Startup, Hyperly

My tech journey spans across 12 years. I've primarily been a backend developer, working with languages like Ruby (Padrino framework), Java (Spring Framework), Scala (Play Framework), and a few others. Along the way, I've built side projects like NoCodeLetters and Breeze, where I got to flex my muscles with Ruby, Serverless, and React.

When my cofounder and I started discussing about Hyperly, our current product, I knew that this was going to be an AI first product that also has to manage a lot of data.

I started evaluating my options, doing some serious research on Python Django, Python Flask, Python FastAPI, and even Java. But Python had a significant advantage with its vast ecosystem. After weighing the pros and cons, I finally decided to go ahead with FastAPI for a few key reasons:

• Asynchronous nature
• Exceptional performance
• Automatic API documentation (goodbye, tedious manual documentation!)
• Usage in High Concurrency Projects (perfect for our needs)

Now, let me be honest, there were moments when I wanted to pull out my hair (especially coming from Ruby), but the growing FastAPI community was a lifesaver. There were always people willing to lend a helping hand, and I'm forever grateful for that.

As Hyperly is primarily a LinkedIn Growth Tool, we had a few basic feature requirements like content generation, scheduling, security, and more. Here's a rundown of the major packages we use:

• We use both Claude (Anthropic) and OpenAI APIs for our AI writer. To call these APIs, we rely on the Async HTTPX package.
• For scheduling those posts, we've got APScheduler 🙇
• And to connect with MongoDB, we've got PyMongo

All of this is hosted on an AWS EC2 instance. We also use SES for sending transactional emails and S3 to save images.

Now, when it comes to the frontend, one thing we learned from NoCodeLetters is how crucial it is to nail the content and SEO game. With that in mind, we're using NextJS for both our landing page and the dashboard.

But that's not all! We've also got a whole suite of other tools that keep us going:

1. Stripe - To collect payments
2. Google Analytics (We are still here 😅)
3. Microsoft Clarity (🤐)
4. Crisp - Customer Support
5. Hubspot - Leads
6. Zoho Mail - It's free
7. Notion - 🛠️

I did try some feature request tools but decided not to use them for the time being.

Resources:

1. FastAPI Tutorial
2. FastAPI Official Documentation

There you have it, folks! That's the tech stack and thought process behind building our AI startup, Hyperly. We are just getting started and I am sure this list will grow.

But do share what other tools we must check out or use!!

🚀