The work is building the systems research runs on, and the research itself: market data, trade reconstruction from broker fills, statistical tests, and the bench of invented traders we check them against.
Research here is reviewed by someone who did not run it, and a result gets one look at held-out data run by someone who does not rule on it. Those separations are the job, not an overhead on it.
A first month is spent reading the pre-registered framework and reproducing a result that is already on the record, including one that failed.