Whenever you're ready, walk me through how you'd approach this.
How do you handle unexpected data or experimental failures and what steps do you take to recover progress?