Inspiration
When i started my startup in Australia, I want to find talent from my university, it is often difficult to do that using linkedin, which is why I though what if RAG + LLMs + Agentic systems could help solve this problem. This is why I started the headhunter project.
What it does
It uses RAG Embedding models , does extensive prompt engineer and grounding for various places around the globe, for eg : we find universities in various countries, so while you search for talent from a city, we usually find folks from the top universities from that city.
How we built it
Using RAG models trained on linkedin data as well as Gemini powered agents that reason over scraped data to display the appropriately.
Challenges we ran into
It is currently bit buggy, for eg : finding phd talent works almost everywhere, but other categories can be a little buggy. But the MVP is ready.
Accomplishments that we're proud of
The accuracy for countries like australia are super accurate and we are extremely proud that accelerators like Techstars are already using this.
What we learned
How important prompt engineering is for projects like these and grounding on real links.
What's next for HeadHunter - A Headhunting tool powered by agents!
Make it super and expand it to all countries and major cities around the world!
Built With
- exa
- express.js
- gemini
- nixpacks
- node.js
- railway
- react
- recharts
- sqlite
- tailwind
Log in or sign up for Devpost to join the conversation.