AI Engineer

☐ Python (Advanced) · Required · 6 hrs

Async, decorators, dataclasses, type hints, packaging. AI engineering requires clean, production-grade Python — not notebook code.

• What is Python? Setup & Your First Program (Python): https://chaduvuko.com/learn/python/what-is-python-setup
• Functions — Defining, Parameters, Return Values (Python): https://chaduvuko.com/learn/python/functions
• Reading & Writing Files (Python): https://chaduvuko.com/learn/python/reading-writing-files
• Exception Handling (Python): https://chaduvuko.com/learn/python/exception-handling

☐ Math for AI · Required · 8 hrs

Linear algebra, probability, statistics, calculus (for understanding gradients). You need enough to read papers and debug loss curves.

• Vectors, Matrices and Tensors (AI & Machine Learning): https://chaduvuko.com/learn/ai-ml/math-foundations/vectors-matrices-tensors
• Derivatives, Gradients and the Chain Rule (AI & Machine Learning): https://chaduvuko.com/learn/ai-ml/math-foundations/derivatives-and-gradients
• Probability Distributions and Bayes Theorem (AI & Machine Learning): https://chaduvuko.com/learn/ai-ml/math-foundations/probability-distributions

☐ Git & APIs · Required · 3 hrs

Version control and calling REST APIs. LLM providers are all API-first.

• Git and Version Control for Data Projects (Data Engineering): https://chaduvuko.com/learn/data-engineering/git-for-data

☐ Deep Learning · Required · 12 hrs

Neural networks, backpropagation, CNNs, RNNs, attention mechanism. Understand what happens inside the models you deploy.

• Neural Networks from Scratch (AI & Machine Learning): https://chaduvuko.com/learn/ai-ml/deep-learning/neural-networks-from-scratch
• Backpropagation — How Neural Networks Learn (AI & Machine Learning): https://chaduvuko.com/learn/ai-ml/deep-learning/backpropagation
• CNNs — Shopify Product Image Classification (AI & Machine Learning): https://chaduvuko.com/learn/ai-ml/deep-learning/cnns-image-classification

☐ Transformers & LLMs · Required · 10 hrs

Attention, BERT, GPT, tokenisation, HuggingFace. The architecture behind every modern language model.

• Transformers and Self-Attention (AI & Machine Learning): https://chaduvuko.com/learn/ai-ml/deep-learning/transformers-and-attention
• LLMs — Pretraining, RLHF and Scaling Laws (AI & Machine Learning): https://chaduvuko.com/learn/ai-ml/generative-ai/llms-pretraining-rlhf

☐ Prompt Engineering · Required · 4 hrs

Zero/few-shot, chain-of-thought, system prompts, structured output. The first skill that separates AI engineers from casual API users.

• Prompt Engineering (AI & Machine Learning): https://chaduvuko.com/learn/ai-ml/nlp/prompt-engineering

☐ Vector Databases · Required · 5 hrs

Embeddings, similarity search, Pinecone/Chroma/pgvector. The memory layer for AI applications.

• Dot Product and Similarity (AI & Machine Learning): https://chaduvuko.com/learn/ai-ml/math-foundations/dot-product-similarity
• RAG — Retrieval-Augmented Generation (AI & Machine Learning): https://chaduvuko.com/learn/ai-ml/nlp/rag-retrieval-augmented-generation

☐ RAG Systems · Required · 8 hrs

Retrieval-augmented generation, chunking strategies, re-ranking, hybrid search. RAG is the dominant pattern for enterprise AI.

• RAG — Retrieval-Augmented Generation (AI & Machine Learning): https://chaduvuko.com/learn/ai-ml/nlp/rag-retrieval-augmented-generation
• Advanced RAG (AI & Machine Learning): https://chaduvuko.com/learn/ai-ml/generative-ai/advanced-rag

☐ Fine-tuning & RLHF · Optional · 10 hrs

LoRA, QLoRA, instruction tuning, DPO. When general LLMs aren't enough and you need domain-specific behaviour.

• Fine-Tuning with PEFT — LoRA and Adapters (AI & Machine Learning): https://chaduvuko.com/learn/ai-ml/nlp/peft-lora-adapters
• LLM Fine-Tuning in Practice (AI & Machine Learning): https://chaduvuko.com/learn/ai-ml/generative-ai/llm-fine-tuning
• LLMs — Pretraining, RLHF and Scaling Laws (AI & Machine Learning): https://chaduvuko.com/learn/ai-ml/generative-ai/llms-pretraining-rlhf

☐ Agents & Tool Use · Required · 8 hrs

LangChain/LlamaIndex agents, function calling, tool orchestration, memory. Multi-agent systems are the frontier of production AI.

• LLM Agents and Tool Use (AI & Machine Learning): https://chaduvuko.com/learn/ai-ml/nlp/llm-agents-and-tool-use
• Agents and Tool Use (AI & Machine Learning): https://chaduvuko.com/learn/ai-ml/generative-ai/agents-tool-use

☐ LLM Evaluation · Required · 4 hrs

RAGAS, faithfulness, answer relevancy, hallucination detection. You can't improve what you don't measure.

☐ AI Deployment & MLOps · Required · 8 hrs

FastAPI model serving, Docker, monitoring latency/cost, prompt versioning, A/B testing prompts. Moving from demo to production.

• Model Deployment — FastAPI, Docker, K8s (AI & Machine Learning): https://chaduvuko.com/learn/ai-ml/mlops/model-deployment
• Model Monitoring — Drift Detection (AI & Machine Learning): https://chaduvuko.com/learn/ai-ml/mlops/model-monitoring

☐ Multi-modal AI · Optional · 6 hrs

Vision-language models (GPT-4o, Gemini), image generation (Stable Diffusion), speech AI. Expanding beyond text-only AI.

• Multimodal Models — CLIP and LLaVA (AI & Machine Learning): https://chaduvuko.com/learn/ai-ml/generative-ai/multimodal-models