Applied AI Engineering: From Fundamentals to Edge and Autonomous Agents
This hands-on training equips software engineers with the end-to-end skills needed to integrate modern AI into real-world applications. Moving beyond theory, participants will explore the complete AI lifecycle: mastering core model architectures (from CNNs to LLMs), optimizing networks for edge deployment (NPUs), and leveraging autonomous AI agents to streamline development workflows.
The 3 modules can be offered together or separately (1 module = 5 hours per day; 3 full days of training). The course content can be adapted based on specific needs. Each module can be taken independently. However, for Module 2 (Edge AI), it is highly recommended to complete Module 1 (AI Fundamentals) first.
Prerequisites:
- Basic Python programming skills (ability to read code).
- Experience with a Linux environment would be beneficial.
Module 1 - AI Fundamentals:
- History of AI
- Convolutional Neural Network - CNN
- Training & Inference
- Attention, Transformer & LLM
- Vision Transformer - ViT
- Frameworks & Tools
Practice:
(1) Train and Infer a CNN with Tensorflow and Pytorch (2) LLM / VLM inference with HuggingFace and llama.cpp
Module 2 - Edge AI:
- Compression and Optimization Techniques
- Neural Processing Units - NPUs
- From Pytorch training to NPU deployment
- Pre- & Post-Processing
Practice:
(1) Model Conversion with ONNX (2) Model Compilation for NPU
Module 3 - Agentic AI:
- Agent and LLM
- Context Engineering
- MCP (Model Context Protocol)
- Memory
- AGENTS.md
- Skills
- Security
- AI Governance
Practice:
AI agents for your development processes, powered by Virtual Engineer. (https://github.com/savoirfairelinux/virtual-engineer)