Did the Genentech OA earlier this year while I was passively looking. Sharing because I couldn't find much recent info and the older posts were vague.
Platform: HackerRank. Two coding problems, 90 minutes. You get to pick your language. I did Python.
Difficulty: Both were medium on the LeetCode scale. One was a classic array/sliding-window problem dressed up in biotech language (I genuinely laughed a little). The other was a graph traversal problem. Nothing hard. No DP, no segment trees, nothing that requires knowing some obscure trick. If you're LC medium-comfortable you'll be fine.
No system design or SQL in the OA. That comes later in the process if you pass.
Time pressure: 90 minutes for two mediums felt very reasonable. I finished in about 55 minutes and spent the rest cleaning up edge cases. They do run hidden test cases so don't just get it working for the examples.
One catch: there was a short written section at the end, two or three open-ended questions like "describe your approach" and "what would you change if the input was 10x larger." I almost skipped these because I was relieved the coding was done. Don't skip them. I'm guessing they look at them.
After the OA I heard back in about 8 business days. For a pharma company that is honestly fast. Recruiter call was next.
I'm a mid-level backend SWE, 6 YOE. Role was in their gRED (Research and Early Development) software group.