Data science in auto insurance pricing is heavily regulated. Therefore, the ML technique used in pricing is very restricted, which is less appealing if you prefer the type of work in "real" tech company . I agree with the previous post that management in ds pricing with very limited experience got promoted so fast. What's worse is the management would micro-manage every detail of your project and frequently give bad instructions you must follow. Definitely a very bad working culture for ds. Telematics ds is probably a different story.