Skip to content

Latest commit

 

History

History
61 lines (43 loc) · 2.57 KB

File metadata and controls

61 lines (43 loc) · 2.57 KB

🎙️ LOQI Demo Script (Website Walkthrough)

Goal: Use the website as your slide deck. Smooth scrolling, confident pauses.


1. The Hero Section (Start Here)

[SCREEN: Top of index.html. The "NEXT GEN CREDIT SCORING" Header matches your opening.]

  • "Good afternoon. We are Team LOQI. This is the future of Credit Risk."
  • "We have built a production-grade Quantum Pipeline on the IQM Garnet Processor."

2. The Problem (The "In-Between" Transition)

[ACTION: Scroll down SLOWLY from Hero, stop JUST BEFORE 'Key Achievements'. Talk to the audience.]

  • "But before we show the results, let's talk about the Problem we faced."
  • "The Credit Crunch:"
    • "Real Quantum Hardware is expensive and slow."
    • "Submitting 2,000 circuits one-by-one would take hours and bankrupt our API budget."
    • "Most teams just run on simulators and call it a day."
    • "We refused to do that."
    • "We built a Batch Execution Engine that reduced API overhead by 99%, making real hardware training possible."

3. Key Achievements (Metrics)

[ACTION: Scroll to the 'Metrics Dashboard' section.]

  • "And that engineering breakthrough led to these numbers:"
  • [POINT to 99% Card]: "Here is that 99% efficiency gain. We solved the bottleneck."
  • [POINT to 89% Card]: "Which allowed us to achieve Proof of Learning on real hardware, reducing loss by 89%."
  • [POINT to 71.5% Card]: "And importantly, we validated a 71.5% accuracy on real hardware using a parallel validation track."

4. The "Two-Track" Benchmark

[ACTION: Scroll to 'Two Tracks, One Goal' Battle Cards.]

  • "We didn't just guess. We ran a Two-Track strategy:"

[HOVER over Track A (Red Card)]

  • "Track A (LOQI) was about Production."
  • "We optimized for the IQM Garnet chip (20 qubits) using our batching engine."
  • "This proved the engineering viability."

[HOVER over Track B (Green Card)]

  • "Track B (Validation) was about Science."
  • "We used Qrisp to solve the Class Imbalance problem."
  • "We sacrificed some accuracy to quadruple our recall. This model actually catches bad loans."

5. Visualizations & Conclusion

[ACTION: Scroll to 'Interactive Visualizations' - Click 'Hardware Trajectory']

  • "Finally, the proof is in the data."
  • "This blue line is our model learning on the Quantum Chip."
  • "In just 24 hours, we delivered a solved engineering problem, verified hardware execution, and a business-ready model."
  • "We are ready to scale. Thank you."