🗺️ AI Learning Paths — DataGate.ch
Your structured journey from beginner to expert. Choose a path, follow the modules, master AI.
Start from zero. Build a solid foundation in artificial intelligence, machine learning, and how modern AI systems work.
1
What Is AI? A Practitioner’s Definition
History, types of AI, and what AI can and cannot do today
2
Machine Learning Basics: Supervised, Unsupervised, Reinforcement Learning
The three paradigms every AI practitioner must understand
3
Neural Networks Demystified: From Perceptrons to Transformers
How neural networks learn, backpropagation, and the architecture behind GPT
4
Large Language Models: How They Work
Tokenization, attention mechanisms, training, and inference
5
Prompt Engineering Fundamentals
Zero-shot, few-shot, chain-of-thought, and system prompts
6
Evaluating AI: Benchmarks, Metrics, and Testing
How to measure AI quality: accuracy, perplexity, human evaluation
Master text AI: from embeddings and RAG to building production NLP systems.
1
Text Representations: Word Embeddings to Contextual Embeddings
Word2Vec, BERT embeddings, and modern sentence transformers
2
Retrieval-Augmented Generation (RAG): Complete Guide
Vector databases, chunking strategies, hybrid search, and reranking
3
Advanced RAG Patterns: HyDE, Self-RAG, Corrective RAG
Production RAG techniques that dramatically improve quality
4
Fine-Tuning LLMs for Text Classification and Extraction
LoRA, QLoRA, and when to fine-tune vs. use RAG
5
NLP Production Systems: Latency, Cost, and Reliability
Deploying NLP at scale: batching, caching, model distillation
Understand how AI sees: image classification, object detection, segmentation, and multimodal models.
1
Image Classification: CNNs, Vision Transformers, and Transfer Learning
From ResNet to ViT: how computers learn to categorize images
2
Object Detection and Segmentation
YOLO, SAM, and detecting/segmenting objects in images
3
Multimodal AI: Vision + Language Models
CLIP, LLaVA, GPT-4V: models that understand images and text
4
Generative Image Models: Diffusion, GANs, and Image Synthesis
Stable Diffusion, DALL-E, and the math behind image generation
Build AI agents that plan, use tools, and act autonomously. The most in-demand AI skill in 2026.
1
Agent Architecture: ReAct, Plan-and-Execute, and Reflexion
Core patterns for building reasoning agents
2
Tool Use and Function Calling
Connecting LLMs to APIs, databases, and code execution
3
Agent Memory Systems: Vector DBs vs Knowledge Graphs
Short-term, long-term, and episodic memory for agents
4
Multi-Agent Coordination Patterns
CrewAI, AutoGen, LangGraph: orchestrating multiple agents
5
Agent Security: Prompt Injection, Guardrails, and Sandboxing
Protecting your agents from adversarial attacks
6
Agent Observability and Evaluation
Tracing, monitoring, and measuring agent performance in production
Production AI engineering: deploy, monitor, optimize, and scale AI systems in the real world.
1
Model Serving Architecture: vLLM, TGI, and Triton
High-throughput LLM serving with batching, KV-cache, and quantization
2
GPU Optimization for AI Workloads
Tensor parallelism, pipeline parallelism, and memory optimization
3
LLM Inference Optimization: Quantization, Speculative Decoding, Batching
Reducing inference cost by 10x without quality loss
4
Fine-Tuning at Scale: LoRA, QLoRA, DPO, and GRPO
Parameter-efficient fine-tuning and RLHF pipelines
5
AI Infrastructure Cost Management
Token budgeting, model routing, and cloud cost optimization
6
Production Monitoring: Drift Detection, A/B Testing, and Alerting
Keeping AI systems reliable after deployment
7
AI Governance: EU AI Act, NIST RMF, and Compliance
Regulatory requirements for AI systems in production
Each module links to existing DataGate.ch content. Learning paths are regularly updated as new content is published. Browse the full content hub →
Schreibe einen Kommentar