I agree that AI hiring requires looking beyond just coding skills. For larger or ongoing projects, businesses also need to consider experience with cloud infrastructure, data engineering, APIs, security, testing, and scaling AI applications.

I think the freelancer vs. dedicated team decision largely depends on project complexity and how long the business expects to maintain the product. For projects involving multiple technologies and continuous development, a dedicated software development team can provide access to different specialists under one team while making it easier to scale development resources as requirements evolve.

Another important factor is whether the project actually needs a custom-trained model. In many cases, existing AI APIs and models may be sufficient, while custom development makes more sense when there are specific data, performance, or business requirements.

It would be interesting to hear from others about when they chose APIs versus custom AI development.