The Human Funnel

Product Idea
Adding Job Profile
Batch processing in 3 stages
Detailed candidate report
Detailed candidate report

Inspiration

Recruiters and hiring teams are overwhelmed by thousands of applications per job post. Online assessments and AI-generated resumes with copy-paste projects make it harder to find genuine talent, and even assignment-based hiring attracts a flood of applications because of AI tools. We wanted to create a fair and scalable system to surface the best candidates quickly and reliably.

What it does

The Human Funnel is a multi-layer AI-driven hiring pipeline that filters tens of thousands of applicants down to the top 100, or even a smaller number, where on-site or more proctored and detailed interviews can take place catering for company alignment and needs. It combines resume parsing, LinkedIn and GitHub analysis for authentic contributions, real projects, qualities and spontaneous phone-call interviews to prevent cheating. Each layer assigns weights based on the job profile, producing a ranked shortlist of the most qualified candidates.

Application Flow Diagram

[

Scoring Algorithm:

Composite Score = (Resume Score × Resume Weight + 
                  LinkedIn Score × LinkedIn Weight + 
                  GitHub Score × GitHub Weight + 
                  Interview Score × Interview Weight) / Total Available Weight

How we built it

We started with a very broad idea in mind. With only a superficial blueprint of what we wanted to build. We leveraged Kiro to break it into clear requirements, implementation plans which were then converted to specifications and manageable subtasks by the power of Kiro IDE. Kiro generated implementation plans and comprehensive test cases, ensuring minimal errors. Its vibe-coding mode let us add small features quickly, and with agent hooks, our documentation stayed up to date while we focused on building core functionality.

Challenges we ran into

We had to design a robust scoring system that balanced different data sources. Ensuring real-time phone-call interviews were seamless and cheat-resistant was also challenging. Integrating multiple parsers and maintaining data accuracy across layers required careful planning and loads of debugging!

Accomplishments that we're proud of

We built a working prototype that successfully filters hundreds of applications and delivers a ranked shortlist. The prototype includes functional parallel parsing of hundreds of resume, functional linkedIn and github parsers and real phone call interviews.

What we learned

Spec-driven development with Kiro significantly accelerates building complex systems. Breaking large problems into smaller tasks with AI-assisted planning improves collaboration and reduces bugs.

What's next for The Human Funnel

We plan to add more data sources like coding platform histories and build dashboards for analytics. We also aim to refine the phone interview system with AI-driven scoring and expand our integrations to make the funnel even more comprehensive. We would like to move ahead from only software role filtering..