Inspiration
Recruiters often waste hours sifting through resumes without clear insight into true candidate fit. We wanted to empower HR teams with an agentic, AI-powered solution that could evaluate, explain, and adapt—without requiring technical expertise or manual sorting.
What it does
JobHatch Enterprise is an all-in-one recruiting intelligence platform that empowers HR professionals to make faster, smarter hiring decisions. Recruiters upload a job description and up to 5 candidate resumes, and the AI agent automatically:
- Extracts job intent using vector embeddings
- Parses resumes to extract structured skills and experience
- Ranks candidates based on vector similarity and LLM-generated match explanations
- Cross-validates profiles with real-time web intelligence (LinkedIn, GitHub, portfolios, etc.)
- Collects interviewer feedback to inform future recommendations
- Learns team preferences over time to continuously improve matching accuracy
How we built it
Figma: Designed the frontend experience Next.js + Supabase: Resume and JD upload, feedback capture LlamaIndex + OpenAI Embedding: JD and resume vectorization Chroma: Vector database for candidate profiles ZeroEntropy: Feedback loop analysis Gemini 2.5 Flash: Multi-step reasoning and explanation (DeepMind) Arcade: Visual walkthrough demo for ToB showcase
Challenges we ran into
Parsing diverse PDF formats reliably
Balancing LLM latency with fast real-time scoring
Designing a scalable and interpretable feedback loop
Accomplishments that we're proud of
- Built a fully working AI agent loop within 5 hours
- Achieved interpretable candidate scoring with embedded GPT explanations
- Delivered cross-validation capability with live internet signals
- Created a ToB-ready deployment flow with UI/UX clarity
What we learned
-Agent-based architecture improves modularity in HR workflows
- Combining vector search + LLMs makes HR evaluations powerful and explainable
- Cross-validation builds trust in automated decision-making
What's next for JobHatch Enterprise
- Scale resume handling: Expand the system to process 10 or more candidate resumes per job and support comparative matching across multiple job roles.
- Enable talent mapping: Introduce a candidate mapping system that visualizes applicant strengths, gaps, and role fit across different departments or positions.
- Pilot launch with partners: Deploy the platform with HR teams at partner companies during the upcoming fall recruitment season, collecting feedback to refine features and improve accuracy.