|
所在平台: Udemy |
课程主页: https://www.udemy.com/course/generative-ai-for-software-engineers-developers/
课程评论：没有评论
课程名称：为软件工程师和开发人员准备的生成性人工智能 课程概述： “为软件工程师和开发人员准备的生成性人工智能”课程旨在赋予现代开发人员在软件开发生命周期中集成前沿AI工具的技能。课程首先建立一个坚实的基础，通过现实世界的例子来解释生成性人工智能的概念，随后探索生成性人工智能的工作原理，包括Transformer和扩散模型。学习者将明确区分软件上下文中的预测性AI与生成性AI，理解生成性人工智能如何改变代码生成、错误修复、文档编写、DevOps自动化和架构设计等任务。 课程中包含的实际例子涉及使用GPT-4、Claude 3、Codex、Gemini 1.5和CodeLlama。深入讲解大语言模型（LLM）的架构，介绍Transformer网络、自注意力机制，以及标记化、上下文窗口和模型局限性等概念。学习者将比较微调与上下文学习，并研究专门的代码LLM，如Codex、StarCoder、CodeGen和AlphaCode。 实践课程介绍了如何通过OpenAI、Hugging Face和Anthropic访问模型API。此外，课程还构建了提示工程的专业知识，涵盖有效原则、零-shot、one-shot、few-shot提示、思维链（CoT）和思维树（ToT）技术，以及创建可重用的提示模板。 在应用设计方面，学习者将探讨AI建议的架构模式，生成实体关系图、时序图，进行架构权衡分析，并评估技术栈。实用的编码模块教授多文件代码生成、类/模块/函数创建、使用SOLID/DRY原则进行代码重构、添加文档以及生成性AI驱动的PR审核。 课程的后续部分聚焦于静态分析、错误检测、单元/集成测试、Dockerfile/Kubernetes清单生成、基础设施即代码(IaC)脚本编写，以及使用Prometheus和Grafana的监控设置。安全性也通过安全代码生成、威胁建模提示、合规性自动化（SOC2、HIPAA、GDPR）以及在静态应用安全测试(SAST)和动态应用安全测试(DAST)中的AI集成得到体现。 最后，学习者将获得一个超过1000个专门设计的提示集合，以提高使用生成性人工智能进行软件工程的生产力。
The "Generative AI for Software Engineers & Developers" course is designed to empower modern developers with the skills to integrate cutting-edge AI tools into the software development lifecycle. Beginning with a solid foundation, the course explains What is Generative AI through real-world examples, followed by an exploration of how GenAI works, covering Transformer and Diffusion models. Learners will clearly differentiate predictive AI from generative AI in software contexts, understanding how GenAI transforms tasks like code generation, bug fixing, documentation, DevOps automation, and architecture design. Practical examples include working with GPT-4, Claude 3, Codex, Gemini 1.5, and CodeLlama.A deep dive into the architecture of LLMs explains Transformer Networks and Self-Attention, alongside concepts like tokenization, context windows, and model limitations. Learners will compare fine-tuning vs in-context learning and study specialized code LLMs like Codex, StarCoder, CodeGen, and AlphaCode. Hands-on sessions introduce accessing model APIs via OpenAI, Hugging Face, and Anthropic. The course also builds expertise in prompt engineering covering effective principles, zero-shot, one-shot, few-shot prompting, Chain of Thought (CoT) and Tree of Thought (ToT) techniques, and creating reusable prompt templates.Moving into application design, learners will explore AI-suggested architecture patterns, generate ER diagrams, sequence diagrams, conduct architectural trade-off analyses, and evaluate technology stacks. Practical coding modules teach multi-file code generation, class/module/function creation, code refactoring using SOLID/DRY principles, adding documentation, and GenAI-driven PR reviews. Further sections focus on static analysis, bug detection, unit/integration testing, Dockerfile/Kubernetes manifest generation, IaC scripting, and monitoring setup using Prometheus and Grafana.Security is integrated through secure code generation, threat modeling prompts, compliance automation (SOC2, HIPAA, GDPR), and AI in SAST/DAST. Finally, learners receive access to a curated 1000+ prompts specifically designed for boosting software engineering productivity with Generative AI.
