# Tutorial Script: Getting Started with TrustformeRS Mobile
**Duration**: 5 minutes
**Target Audience**: Mobile developers new to TrustformeRS
**Prerequisites**: Basic iOS/Android development knowledge
## Script Overview
This tutorial introduces developers to TrustformeRS mobile deployment, covering installation, basic setup, and running your first on-device inference example.
---
## Opening (0:00 - 0:30)
**[Visual: TrustformeRS logo and mobile devices]**
**Narrator**:
"Welcome to TrustformeRS Mobile! I'm excited to show you how to bring state-of-the-art transformer models directly to iOS and Android devices. In just 5 minutes, you'll learn how to set up TrustformeRS and run your first on-device AI inference."
**[Visual: Split screen showing iPhone and Android device]**
"TrustformeRS Mobile enables privacy-preserving, lightning-fast inference without internet connectivity. Let's get started!"
---
## Section 1: What is TrustformeRS Mobile? (0:30 - 1:30)
**[Visual: Architecture diagram showing on-device inference]**
**Narrator**:
"TrustformeRS Mobile is a comprehensive framework that brings transformer models to mobile devices. Unlike cloud-based solutions, everything runs locally on your users' devices."
**[Visual: Feature highlights with icons]**
"Key features include:
- **Native iOS and Android support** with Swift and Kotlin APIs
- **Hardware acceleration** using Core ML, Metal, NNAPI, and Vulkan
- **Model optimization** with automatic quantization and compression
- **Privacy-first design** - no data leaves the device
- **Production-ready** with memory management and battery optimization"
**[Visual: Performance comparison chart]**
"You get enterprise-grade performance while maintaining complete data privacy."
---
## Section 2: Installation and Setup (1:30 - 2:45)
**[Visual: IDE screen showing project setup]**
**Narrator**:
"Let's set up TrustformeRS in your mobile project. I'll show you both iOS and Android setup."
### iOS Setup
**[Visual: Xcode with Swift Package Manager]**
"For iOS, open Xcode and add TrustformersKit via Swift Package Manager. In your project settings, click 'Package Dependencies' and add:"
**[Text overlay: https://github.com/trustformers/trustformers-mobile]**
```swift
// Add to your imports
import TrustformersKit
// Initialize the engine
let engine = TFKInferenceEngine()
```
### Android Setup
**[Visual: Android Studio with Gradle files]**
"For Android, add TrustformeRS to your `build.gradle`:"
**[Code display]**
```kotlin
dependencies {
implementation 'com.trustformers:trustformers-mobile:1.0.0'
}
```
```kotlin
// Initialize in your Activity or Fragment
val engine = TrustformersEngine(this)
```
**[Visual: Dependency sync in progress]**
"Sync your project, and you're ready to go!"
---
## Section 3: Your First Inference (2:45 - 4:15)
**[Visual: Code editor with syntax highlighting]**
**Narrator**:
"Now let's run your first on-device inference. We'll use a pre-trained BERT model for text classification."
### iOS Example
**[Visual: Swift code being typed]**
```swift
class ViewController: UIViewController {
let engine = TFKInferenceEngine()
override func viewDidLoad() {
super.viewDidLoad()
runInference()
}
func runInference() {
// Load a pre-trained model
engine.loadModel("bert-base-uncased") { result in
switch result {
case .success:
self.classifyText()
case .failure(let error):
print("Model loading failed: \(error)")
}
}
}
func classifyText() {
let text = "TrustformeRS makes mobile AI easy!"
engine.classify(text: text) { result in
switch result {
case .success(let classification):
print("Sentiment: \(classification.label)")
print("Confidence: \(classification.confidence)")
case .failure(let error):
print("Classification failed: \(error)")
}
}
}
}
```
### Android Example
**[Visual: Kotlin code being typed]**
```kotlin
class MainActivity : AppCompatActivity() {
private lateinit var engine: TrustformersEngine
override fun onCreate(savedInstanceState: Bundle?) {
super.onCreate(savedInstanceState)
setContentView(R.layout.activity_main)
engine = TrustformersEngine(this)
runInference()
}
private fun runInference() {
// Load a pre-trained model
engine.loadModel("bert-base-uncased") { result ->
when (result) {
is Result.Success -> classifyText()
is Result.Error -> Log.e("AI", "Model loading failed: ${result.error}")
}
}
}
private fun classifyText() {
val text = "TrustformeRS makes mobile AI easy!"
engine.classify(text) { result ->
when (result) {
is Result.Success -> {
Log.i("AI", "Sentiment: ${result.data.label}")
Log.i("AI", "Confidence: ${result.data.confidence}")
}
is Result.Error -> Log.e("AI", "Classification failed: ${result.error}")
}
}
}
}
```
**[Visual: App running on device/simulator]**
"Run your app, and you'll see the classification results in your console. The model runs entirely on-device!"
---
## Section 4: Performance and Optimization (4:15 - 4:45)
**[Visual: Performance monitoring dashboard]**
**Narrator**:
"TrustformeRS automatically optimizes for your target devices. Let's see what's happening under the hood."
**[Visual: Code showing configuration options]**
```swift
// iOS: Enable hardware acceleration
let config = TFKModelConfig()
config.useNeuralEngine = true
config.useMetal = true
config.precision = .fp16
engine.loadModel("bert-base-uncased", config: config)
```
```kotlin
// Android: Configure hardware acceleration
val config = TrustformersConfig.Builder()
.enableNNAPI(true)
.enableGPU(true)
.setPrecision(Precision.FP16)
.build()
engine.loadModel("bert-base-uncased", config)
```
**[Visual: Performance metrics overlay]**
"This enables Core ML Neural Engine on iOS and NNAPI on Android for maximum performance while maintaining accuracy."
---
## Conclusion and Next Steps (4:45 - 5:00)
**[Visual: Next tutorial previews]**
**Narrator**:
"Congratulations! You've just run your first on-device AI inference with TrustformeRS. In our next tutorials, we'll dive deeper into:"
**[Text overlay with bullet points]**
- Advanced model optimization techniques
- Custom model deployment
- Production best practices
- Performance profiling and debugging
**[Visual: TrustformeRS community links]**
"Check out the documentation, join our community Discord, and don't forget to star us on GitHub. Thanks for watching, and happy coding!"
---
## Technical Notes
### Recording Instructions
1. **Screen Setup**: Use 1920x1080 resolution with high-contrast IDE theme
2. **Code Typing**: Type at moderate pace (2-3 characters per second)
3. **Device Recording**: Use device screen recording for demo sections
4. **Audio**: Clear, professional narration with consistent pacing
### Required Assets
- TrustformeRS logo and branding assets
- IDE screenshots and recordings (Xcode, Android Studio)
- Device recordings (iPhone, Android)
- Performance dashboard mockups
- Community resource graphics
### Post-Production
- Add closed captions for accessibility
- Include chapter markers at section breaks
- Add animated callouts for important code sections
- Include download links in video description
- Create thumbnail with clear branding
### Code Repository
All code examples should be available in a companion GitHub repository:
- `/ios-getting-started/` - Complete iOS project
- `/android-getting-started/` - Complete Android project
- `README.md` - Setup instructions and requirements
- `requirements.txt` - Dependencies and versions
### Testing Checklist
- [ ] Code examples compile and run successfully
- [ ] All dependencies are available and version-pinned
- [ ] Performance claims are accurate and measurable
- [ ] Accessibility features work correctly
- [ ] Links and resources are valid and up-to-date
---
*Tutorial Script Version: 1.0*
*Last Updated: 2025-07-16*
*Estimated Recording Time: 6-8 hours (including setup and editing)*