Don't read papers start-to-finish. Use a multi-pass approach.
Pass ( min): Title, abstract, figures, conclusions. Decide if it's worth reading.
Pass ( min): Introduction, method overview, experiments. Understand the main idea.
Pass (+ hours): Deep dive. Understand every equation. Reproduce if needed.
What to extract:
- Problem and motivation
- Key insight/contribution
- Method summary
- Results and limitations
Interview tip: For each paper, prepare a -minute summary and criticism.