Alibaba ACP-LLM Engineer Certification Exam Syllabus

• Basic API parameters, including model, temperature, and top_p;
• Batch and streaming generation;
• Message structure and conversation history

• Prompt frameworks, including prompt elements, separators, and templates
• The role of the system prompt

- Using LLMs for various tasks

• Common LLM use cases
• Developing applications with LLMs for tasks like batch intent classification, document review, and automatic revision

• Core RAG components: file parsing, text chunking, retrieval, and reranking
• RAG retrieval optimization: sentence window retrieval and auto-merging retrieval

- Continuous RAG optimization

• Practical RAG optimization techniques, including text parsing, title rewriting, and table content enhancement

- Automated evaluation of RAG

• The RAGAS framework
• RAG system evaluation methods

• The purpose, prerequisites, basic steps, and common algorithms for fine-tuning

- Fine-tuning experiments and evaluation

• Dataset construction, parameter settings, and model evaluation for finetuning

• How agents work
• Generating complex workflow and multi-agent applications

- Building complex AI applications

• Hands-on practice with Alibaba Cloud AI solutions
• AI applications in healthcare, education, and entertainment

• LLM deployment with vLLM
• Publishing AI assistants with Function Compute (FC)

- Production considerations for LLM applications

• Balancing performance and costs
• Improving application stability
• Ensuring security and compliance