Generative Engineer Associate (300)

Key skills

• Knowledge of GenAI: Understanding the principles and underlying technologies of generative AI, including both text and image models, as well as the core
foundational models (FMs) used for generating content.
• Model Training and Optimization: Comprehension of basic training and finetuning processes for generative models, coupled with knowledge of advanced
techniques such as RAGs and agents to enhance model performance.
• Prompt Engineering Proficiency: Craft effective prompts to optimize the output of GenAI models, ensuring improved accuracy and alignment with user
expectations.
• Identification of AI Solutions: Capability to describe and recommend suitable generative AI solutions and services provided by Alibaba Cloud, such as Qwen
and Model Studio.

Course outline

Course introduction
• Roadmap/agenda
• What the Generative AI Engineer role is about
• Pre-course survey

Module 1: Generative AI Basics
• Introduction to Generative AI Concepts
o What is GenAI
§ Development history of GenAI
§ GenAI vs traditional AI
o Key concepts in GenAI
o Use cases and industry applications
§ Logistics, education, healthcare, etc.
o Core foundational models
§ GANs, VAEs, and Transformers

Module 2: Foundational Technologies Behind AI
• Foundational Technologies Behind Today’s AI Capabilities
o What is machine learning
§ Understanding machine learning
§ Training techniques (supervised, unsupervised, and
reinforcement learning)
§ Types of ML algorithms and their applications
o What is deep learning
§ Biomimicry in deep learning
§ Understanding RNNs and CNNs
o What is NLP
§ How machines can understand language
§ The NLP development pipeline
§ NLP at work
• Tokenization, embeddings, and language models

Module 3: Alibaba Cloud GenAI Offerings
• Introduction to Qwen
o What is Qwen
o Model capabilities and performance
o Qwen model offerings
• Deploying Models on Alibaba Cloud
o Introduction to GPU-accelerated ECS instances
o Introduction to PAI
o Introduction to ComputeNest
• Model Studio Fundamentals
o What is Model Studio
o Concepts & components
§ System prompts
§ Multi-model comparison
§ Agent and RAG creation

Module 4: Generative Model Training & Optimization
• Understanding Language Models
o Evolution of language models
o Scaling laws vs performance
• Pre-training Generative Models
o Data collection & preparation
o Model architectures
o Tuning techniques (instruction and alignment tuning)
• Generative Model Optimization
o Understanding augmentation techniques
o Understanding retrieval augmented generation systems (RAGs)
o Understanding agents
§ Multi-agent workflows

Module 5: Interacting Effectively with Generative Models
• Prompt Engineering Fundamentals
o Understanding how generative models “think”
o What is prompt engineering
o The “genius in a room” analogy
o Crafting prompts:
§ The five elements of a good prompt
§ “Training” LLMs through effective prompting
o Basic prompting techniques
§ Instructions
§ Examples (few-shot)
§ Role-play
§ Chain-of-thought