💥 Launching promotions.fyi

Plus, the tech stack behind Taro

We just launched promotions.fyi: company-specific guides with vetted insights about performance review and promotion. Join the waitlist today for early access.

Yes, I’m very excited that the domain was available 😊 We’re hard at work to make this jam-packed with value for software engineers:

• We’ll cover the basics: when performance review happens, rating breakdowns at each level, and the promotion process

• We’ll also share insights for each company, with advice from engineers at various levels

We’ve talked to hundreds of engineers at target companies already, so I’m really excited to bring this to life.

If you have thoughts on what you’d like to see, or if you want to contribute from your company’s perspective, please reach out.

I’ve gotten a ton of questions about the tech stack for Taro. I’ll probably make a video about this on YouTube, but here’s the gist of it.

First, the guiding principle for Taro is that we care more about users than technology. We’re happy to use boring tech since our priority is to deliver value as quickly as possible.

• Web app: TypeScript, Next.js, Tailwind (deployed with Vercel)

• Android app: Kotlin + XML layouts

• iOS app: Swift UI

• We use Firestore for the database and Firebase Storage for images and videos

• We heavily use Firebase triggers for things like notifications and data synchronization across the NoSQL data store. Our backend is written in TypeScript.

• We have a bunch of internal dashboards: tracking referrals, managing events, and sending out notifications. We initially built these from scratch, but we recently discovered the magic of low-code tools (in particular, Retool)

• Stripe runs our financial infrastructure (payment and billing)

• We use Posthog extensively for analytics

• We use loops.so for sending both transactional and marketing emails

One thing worth calling out: we don’t worry much about compute efficiency. We have 25K active users on Taro. Hopefully that number will double or triple soon, but even then, we are a far cry from social media scale (the typical size for an experiment at Facebook was a few million users). Because we’re small, we focus on developer time instead of machine time.

For example, a new response in Taro’s forum will spawn multiple additional document updates in a fairly inefficient way. We could improve this, but we don’t need to. Taro will succeed based on how much we help engineers in their career; at this stage, it would be foolish to optimize our Google Cloud bill.

---

Unless you’ve been living under a rock, you’ve probably heard of, and perhaps tried, the generative AI chatbots from OpenAI (Chat GPT) and Google (Bard AI). Unlike web3, the hype here is justified:

Hope to see you at one of our upcoming events soon!

Rahul & Alex

joinTaro.com