I recently had one of my clients say they needed a full time Oleg Nikolaev just to do AI research.

I completely agree, everyone needs an Oleg, but I wanted to dive a little deeper on that.

The AI landscape is changing very rapidly for a host of reasons:

1) New models come out daily, and even frontier models like Claude or ChatGPT are released every couple of months (and apparently revoked now too).

2) New open source libraries and frameworks are releasing and updating frequently.

3) With new models (and even with existing models) their performance can vary significantly, even when using the same skills or the same workflows.

4) It is very easy for companies training AI models to inflate their benchmark scores (and they are incentivized to do so), which can make it difficult to evaluate new models.

These reasons are just the tip of the iceberg.

I bring this up because I am seeing this growing trend for companies to expect their software engineering teams to be experts on AI, and to stay up to date on such a fast moving industry.

That is not their job! They already have a day job. If you want to implement cutting edge technology, there is a massive time investment and expertise requirement.

Either dedicate your resources to research, experimentation, benchmarking, testing, etc., or use a firm like Elevate that specializes in this very area, so you and your team can focus on your core business.