Case Study

Moving beyond simple prompt engineering, I am not using LLMs out of the box but rather shifting to AI/LLM engineering, tackling solutions by applying more comprehensive techniques, i.e., customising reference models with algorithms and building necessary logic and architecture for the applications.  

​

API Access in these projects (POE API) 

​

Access to frontier Large Language Models (LLMs) in Hong Kong is limited, mainly due to geographic policy restrictions and concerns regarding censorship and legal compliance. As alternatives, both the POE.COM API and the OPENROUTER API provide gateways for accessing various AI models.I chose the POE API because it is more affordable and accessible on mobile devices. Although it supports fewer models than OPENROUTER, it meets my requirements. Not all statements will be compatible with the OPENAI syntax, but I managed to resolve them using alternative methods. Additionally, the POE API offers limited documentation, so additional research may be necessary. 

​

Python

Python remains the top choice for AI and ML projects due to its simplicity, and its extensive libraries and frameworks make it more effective and scalable to use as the main programming language for AI algorithms.  

 

​