β REST API with Axum
GET /health- Health checkGET /api/stats- System statisticsPOST /api/search- Semantic searchPOST /api/upload- Upload embeddingsGET /api/load- Load dataset from file
β Beautiful Web Interface
- Real-time stats dashboard
- Interactive search with example queries
- Upload embeddings via textarea or file path
- Beautiful gradient design
- Sub-millisecond query time display
- Mobile-responsive layout
Start it: cargo run --release -p vectro_cli -- serve --port 8080
β
Enhanced Demo (demo_enhanced.sh)
- Colored terminal output
- Progress indicators
- Step-by-step walkthrough
- Semantic search examples
- Web server integration
- Perfect for screen recording
β
Quick Demo (demo_quick.sh)
- 30-second teaser
- Shows compression + search
- Social media friendly
β
Original Demo (demo.sh)
- Classic comprehensive demo
- All core features
β
Themed Embeddings (scripts/generate_themed_embeddings.py)
- Products (electronics, clothing, food, books, toys, sports)
- Movies (by genre)
- Documents (by topic)
- Mixed datasets
- Configurable dimensions and count
- Semantic clustering for realistic demos
β
Random Embeddings (scripts/generate_embeddings.py)
- Simple random vectors
- Good for testing
β VIDEO_DEMO.md
- Complete recording guide
- 5-7 minute demo script
- Pre-recording checklist
- Visual tips
- Post-production guide
β QUICKSTART_VIDEO.md
- Quick start options
- API reference
- Demo flow templates
- 3-minute demo structure
β Updated README.md
- Prominent web UI features
- REST API documentation
- Enhanced quick start
# Follow the complete guide
cat VIDEO_DEMO.md
# Run the enhanced demo script
./demo_enhanced.shThis shows:
- Data generation with semantic meaning
- Streaming compression
- Quantization (75% reduction!)
- Semantic search with real results
- Web UI with live demo
./demo_quick.shPerfect for: Twitter, LinkedIn, Instagram stories
# Generate data
python3 scripts/generate_themed_embeddings.py --count 1000 --theme products > products.jsonl
# Compress
cargo run --release -p vectro_cli -- compress products.jsonl products.bin
# Start server
cargo run --release -p vectro_cli -- serve --port 8080
# Then record browser interaction:
# 1. Show dashboard
# 2. Load dataset
# 3. Run searches
# 4. Highlight speed- Streaming Compression - Handle datasets larger than RAM
- 75% Size Reduction - Quantization with minimal accuracy loss
- Microsecond Search - Sub-millisecond query times
- Beautiful UI - Production-ready web interface
- REST API - Easy integration
- Rust - Safe and fast
- β Green checkmarks for success
- β Cyan arrows for data/output
- βΉ Yellow info indicators
- Colored progress bars
- Unicode emojis (π, π, π, etc.)
- Gradient purple design
- Real-time stats cards
- Interactive search interface
- Animated loading states
- Beautiful result cards
- Build release version:
cargo build --release - Test all scripts work
- Generate sample data
- Clear terminal history
- Set font size to 18-24pt
- Use dark theme
- Close other apps
- Disable notifications
- 1920x1080 or 1280x720 resolution
- 30 or 60 fps
- Record system audio (for voiceover)
- Test audio levels
- Add title slide
- Add section chapters
- Speed up compilation (2-4x)
- Add end screen with links
- Export in H.264
vectro-plus/
βββ demo_enhanced.sh # Main demo script
βββ demo_quick.sh # 30-second teaser
βββ demo.sh # Original demo
βββ VIDEO_DEMO.md # Complete recording guide
βββ QUICKSTART_VIDEO.md # Quick reference
βββ vectro_cli/
β βββ src/
β β βββ server.rs # Web server implementation
β β βββ main.rs # CLI with serve command
β βββ static/
β βββ index.html # Web UI
βββ scripts/
βββ generate_themed_embeddings.py
βββ generate_embeddings.py
0:00-0:15 - Title + Introduction
- "Vectro+ - High-performance embedding search in Rust"
0:15-1:00 - Compression Demo
- Generate data
- Show streaming compression
- Show quantization
- Display size savings
1:00-2:00 - Search Performance
- Run semantic searches
- Show query times
- Highlight accuracy
2:00-4:00 - Web UI
- Start server
- Load dataset
- Interactive searches
- Show metrics
4:00-4:30 - Benchmarks (optional)
- Show criterion output
- HTML report preview
4:30-5:00 - Wrap-up + Links
- Key benefits
- GitHub link
- Call to action
- Test Everything First - Run through once before recording
- Use Pauses - Let visuals breathe, give viewers time to read
- Show Real Numbers - File sizes, query times, percentages
- Keep It Focused - One feature at a time
- End Strong - Clear call to action (star the repo, try it out, etc.)
"Vectro+ is a high-performance embedding compression and search engine built in Rust. [pause]
It handles datasets larger than RAM through streaming compression, [show demo] reduces storage by 75% with quantization, [show size comparison] and delivers microsecond search times. [show query]
The web interface provides real-time semantic search with a beautiful dashboard. [show UI]
Built in Rust for safety and speed, Vectro+ is perfect for production embedding systems. [show metrics]
Check it out on GitHub and give it a star if you find it useful. Thanks for watching!"
Everything is set up for a great demo video. Choose your approach:
- Comprehensive demo? β
./demo_enhanced.sh - Quick teaser? β
./demo_quick.sh - Web focus? β Start server + browser recording
- Custom flow? β See VIDEO_DEMO.md for full guide
Good luck with your video! π¬
Built with β€οΈ for great demos!