Get in Touch

Course Outline

Foundations of Edge AI and the Nano Banana Ecosystem

  • Defining the core traits of edge-AI workloads
  • Examining Nano Banana’s architecture and features
  • Weighing the merits of edge versus cloud deployment strategies

Ready-ing Models for Edge Environments

  • Selecting models and establishing performance baselines
  • Addressing dependency and compatibility requirements
  • Exporting models to prepare for further optimization

Advanced Model Compression Methods

  • Applying pruning strategies and structural sparsity
  • Utilizing weight sharing to reduce parameters
  • Assessing the impact of compression techniques

Quantization Strategies for Edge Performance

  • Executing post-training quantization methods
  • Implementing quantization-aware training workflows
  • Exploring INT8, FP16, and mixed-precision approaches

Accelerating Inference with Nano Banana

  • Leveraging Nano Banana’s hardware accelerators
  • Integrating ONNX formats with specific hardware backends
  • Benchmarking the performance of accelerated inference

Deploying Models on Edge Devices

  • Embedding models into mobile or embedded applications
  • Configuring runtime settings and establishing monitoring
  • Diagnosing and resolving deployment challenges

Performance Profiling and Strategic Trade-offs

  • Managing latency, throughput, and thermal limits
  • Navigating the trade-offs between accuracy and speed
  • Developing iterative optimization strategies

Best Practices for Sustain Edge-AI Systems

  • Managing versioning and continuous updates
  • Handling model rollbacks and compatibility
  • Ensuring security and system integrity

Course Recap and Future Directions

Requirements

  • A solid grasp of machine learning workflows
  • Proficiency in Python-based model development
  • Knowledge of neural network architectures

Target Audience

  • ML Engineers
  • Data Scientists
  • MLOps Specialists
 14 Hours

Number of participants


Price per participant

Testimonials (1)

Upcoming Courses

Related Categories