This isn't a big problem. I write in Chinese, but I can "train" the AI to translate into English the way I want. I don't even need to know English. All you need are three AIs: A translates, B improves A, then B's improvements are reviewed by A for revision, and finally C verifies and translates back to the original language to see if there are any discrepancies.
The most important thing in AI training is the consistency between personal names and professional names. You can create a table of these and have the AI reconfirm them before each translation.
Also, once you've adjusted your AI's style to your satisfaction, have it generate a self-generated style evaluation standard. This standard represents the result of your AI training. Later, when you need to translate, compare the standard with the professional names and give it to the AI to memorize. This way, you can maintain a consistent translation style.
This takes some time, but the rewards are immense.
The key issue in English translation is the balance between literary and commercial styles. You can discuss this with AI (B) and then produce a second or third version of the translation based on the original. Comparing the differences between the two versions will lead to significant improvement.
You can refer to my work
Translate Chinese into English using the AI I trained
The Winter of Ambition: War for the Throne