Generative AI for software engineers.

I'm diving deep into the world of Generative AIs, and you can expect to see many more posts from me in this domain. While I skipped blockchain technology and Web 3.0, AI has truly captured my interest, and it's poised to have a much more significant impact on our daily lives.

I recently delved into the origins of AI by studying this pivotal [paper](https://arxiv.org/abs/1706.03762v7) and examining this [code](https://github.com/tunz/transformer-pytorch/blob/e7266679f0b32fd99135ea617213f986ceede056/model/transformer.py#L201). However, there's still a lot of ground to cover in understanding how AI and large language models (LLMs) work, as well as how to train them. Besides this deep dive, I've explored how AI can assist me and my team.

From an engineer's perspective, AI can significantly streamline daily workflows. Here are some critical use cases:

♻️ Data Scraping: While there are existing tools for data scraping, every software engineer has faced this task, whether it involves extracting information from the web or parsing documents. LLMs can now assist in automating and simplifying these tasks.

🧬 Code Generation: Requesting an LLM to generate code snippets saves valuable development time.

✔️ Code Review: LLMs excel at offering suggestions, identifying bugs, and providing potential solutions during code reviews.

🤖 Code Assistance: Imagine having a dynamic, interactive version of Stack Overflow where you can describe your coding issue, read the response, and ask follow-up questions for further guidance.

🧑🏫 Understanding Code: LLMs can explain specific code sections, making it easier to grasp complex logic.

🚜 Refactoring: Whether bringing legacy code up to date, simplifying complex code, or generating test cases, AI can be a powerful aid in code improvement.

🧰 Tool Assistance: LLMs can guide you through various tasks, such as using Git or other development tools.

Understanding that AI is designed to assist experienced engineers rather than replace them is crucial. My recommendation to fellow engineers is to invest time in mastering and incorporating these models into your daily workflow.

Another noteworthy aspect of my experience is its assistance in using tools I don't use daily. In the past, I maintained text files for each cli-tool, documenting their command-line parameters and functionalities. However, with the help of LLMs, I can now conveniently look up information about these tools on the fly.

There are even more cases where AI is a good tool assisting you in your day-to-day job.