Engineering @ Clera
A small team, an AI talent agent, and the principles, models, and tools we ship to make it work. Three places to dig in.
How we build
The product, the stack, and what we open-source.
What we're building
Clera is an AI talent agent. It reads a candidate's background, works out which of the 2,000+ open roles at 900+ startups fit, and makes the warm intro to the founder or hiring manager. Under the hood that is a matching engine, a conversational agent that runs across WhatsApp, iMessage, LinkedIn, Slack and web chat, and the tooling the talent team uses all day.
The engineering team is small, and every engineer owns a slice of that end to end.
The stack
TypeScript throughout. Next.js on Vercel, PostgreSQL with Drizzle, Typesense for search, Trigger.dev for background work, and the Vercel AI SDK with Langfuse for tracing so we can swap models without rewriting anything. Boring, well-documented pieces on purpose - the interesting problems are in matching, ranking and conversation quality, not in infrastructure.
What we give back
The developer tooling we build to run our own infrastructure is public on github.com/getclera: trigger-cli, typesense-cli, typesense-terraform and instantly-cli. Every one of them runs in our production environment.
Engineering Culture
The principles that shape how we work - speed, ownership, simplicity, and the bar we hold each other to.
AI
How we use language models, what we don't use them for, and the stack behind our matching.
Open Source
The developer tooling we built for ourselves and ship publicly - CLIs and Terraform modules.