https://www.loom.com/share/9aec1973064e44bfbf105f4cfaaae094?sid=b0950e43-e0a9-4c9e-bbb7-c0b7885a2dbe
What does it mean to be a future-proof engineer? It’s about being a one-person startup—someone who can wear multiple hats like Product Manager, UI/UX Designer, Fullstack Developer, and AI Engineer. With AI tools, you can now operate at 10x speed and build real-world products solo.
This challenge is a great opportunity for students and junior developers to accelerate their careers by showcasing their full-stack capabilities and learning how to use modern AI tooling to solve real-world problems. Join this hackathon to develop these skills while building an AI-powered talent-matching MVP—similar to Juicebox - PeopleGPT or Mercor.
Mercor recently closed a $140M fundraising round (source), with a core focus on matching people to the right roles. People matching is foundational to many everyday platforms like LinkedIn, Indeed, Fiverr—and yes, even Tinder. It’s a deeply impactful problem that spans recruiting, freelancing, and networking.
Use AI to synthesize and generate the data you need in structured formats like JSON. You can ask AI to imitate data objects such as resumes, LinkedIn profiles, GitHub profiles, and more. Structured output is a key concept when working with the OpenAI API, and will make your prompting and downstream applications much easier.
At its core, this is a recommendation system problem. It’s important to get familiar with metrics like Precision@N, NDCG, and others that evaluate ranking quality. Also note that most recommendation systems follow a two-stage pipeline: Retrieve (generate candidates) and Rank (score and order the best matches).
A very challenging problem is how and where to find the ground-truth, you may want to introduce some human (expert) in the loop mechanism. How would you design it?
Coding with AI is a very different experience compared to traditional development. A great way to begin is by asking AI to prototype your idea. Check out the videos below for inspiration—here’s an example where I used Cursor and Claude 3.7 to help design a solution for another Brave project.
Here’s the prompt I used, make sure you set it in the Thinking mode:
I am creating a web app to help companies host job post and self-serve interviews with AI hiring managers.
Create UI prototypes in html of the interfaces for both the hiring managers view and the candidates view.
Start with following views:
hiring manager
- create a job post
- specify the interview flow
- see candidates' performance
candidate
- see the job requirement on job board
- signup and conduct the interview
- receive feedback
Requirements
1. Think about the key features users need
2. Act as a PM to plan the features
3. Act as a designer to design the features
4. Output all prototypes in one HTML file, use open source, modern icons, prefer Shadcn and react lucide
I want the design, not the source code, I will take this design to a developer so he can implement bas on it
https://www.loom.com/share/ec5f73e9248a41bb8c216d7e4933724b?sid=5f1972c4-2ed2-4ade-ba77-2e1504f14a02