Probably best topic for me would probably be how to statistically model quality in manufacturing. Like what is hypothesis testing, what is a gage r&r, what is spc, how do you properly do those things, what are the footguns, how do you interpret the results, what's a spec limit vs a control limit vs action limit?
How can you manipulate the tests for your business case? To be clear on that last one I'm not saying manipulate to lie, but it depends on business. For example your control limits are typically +/- 3 standard deviations. If you run a dark business making plastic forks but to get a engineer to fly out and adjust the machine is a few thousand dollarsand no one will care if the tongs on the fork are 3 thou longer than normal. You may set your control limits at 5 times. If you make the steel beams which hold up bridges where hundreds of people die if they can't hold the forces and you're measuring width of the beams you may go down to 1 or 2 times standard deviation. Yeah you will waste engineer time which is money but it's hundred of people's lives if it fails. Those are different business cases and you shouldn't just default to the standard 3 st.dev in those cases.