Axon: The Cognitive Interview Engine

Next-gen AI-powered video interview platform for data-driven hiring and real-world evaluation.


Inspiration

Hiring today is still dominated by resume screening, keyword matching, and subjective shortlisting. Strong candidates are often rejected before they ever get a chance to speak, while recruiters struggle to evaluate real skills from static resumes.

We wanted to challenge this resume-first mindset. What if hiring focused on real performance instead of paper profiles? With the rise of multimodal AI like Gemini, we built Axon—a platform where candidates are seen, heard, and evaluated on actual ability.

What it does

Axon is a dual-portal ecosystem designed for high-velocity, high-integrity hiring.

For Recruiters:

  • Cognitive Job Architect: Define roles with AI-driven difficulty and custom question counts.
  • Executive Intelligence Overview: A synthesized dashboard showing an Averaged Competency Matrix and Overall Assessment Grade.
  • Integrity Sensor: A real-time compliance tracker that flags "Proxy Interviewing" or cheating markers using Gemini 3 Pro analysis.
  • Hybrid Scheduling: 1-click instant rooms via Jitsi Meet or a "Create & Link" workflow for Google Meet to ensure joining never fails.

For Candidates:

  • High-Fidelity Interview Engine: An immersive recording environment with AI voice (Gemini 2.5 TTS) and synchronized text-to-speech.
  • Resume-Job Validation: Real-time mismatch detection. If a resume doesn't align with core requirements, Gemini explains why immediately.
  • Career Dashboard: Track invitations, live statuses (Shortlisted, Under Review), and upcoming scheduled syncs.

How we built it

We utilized a cutting-edge, serverless AI stack:

  • Frontend: React 19 with a high-end "Glassmorphism" aesthetic and Tailwind CSS.
  • AI Orchestration: Google GenAI SDK (@google/genai) using:
    • Gemini 3 Pro: For complex evaluation, grading, and behavioral reasoning.
    • Gemini 3 Flash: For ultra-fast, high-precision transcription and resume parsing.
    • Gemini 2.5 Flash TTS: To provide a natural, professional interviewer voice.
  • Backend & Security: Firebase (Firestore, Auth, Storage) for real-time data streaming and secure video asset management.

Challenges we ran into

  1. The Google Meet "Closed Loop": Randomly generated Meet links often lead to broken sessions. We solved this by implementing a "Verified Creator" workflow where the recruiter creates a real room and Axon validates the URL.
  2. Auth Synchronization: Solved the "Blank Screen on Login" issue by ensuring the Auth state, User Profile, and initial Job lists resolve in a single, blocking startup sequence.
  3. High-Fidelity Transcription: Standard STT often misses technical jargon. We implemented a protocol using Gemini 3 Flash that ignores background noise to focus purely on technical accuracy.

Accomplishments that we're proud of

  • Zero-Refresh Login: A seamless transition from authentication to dashboard.
  • 100% Automated Grading: A system that evaluates Relevance, Clarity, Technical Depth, Problem Solving, and Integrity without human intervention.
  • Red Flag Compliance: Automated detection of integrity scores under 60% with visual recruiter alerts.

What we learned

  • Multimodal AI works best when text, audio, and video are combined to form a complete candidate profile.
  • Context-aware evaluation drastically improves hiring quality compared to keyword-based ATS systems.
  • AI should assist humans, providing data-driven "Executive Insights" rather than replacing the final human decision.

What's next for Axon

  • VEO 3.1 Integration: Using video generation to create custom AI avatars for the interviewers.
  • Bias-Detection Overlay: An AI layer that flags potentially biased language in recruiter job descriptions.
  • Deep-Link Integrations: One-click calendar invites (ICS) for scheduled interviews.

Axon: Making hiring fair, intelligent, and performance-driven.

Built With

Share this project:

Updates