- seed joint optimizer in fast waypoints → make it more targeted
- use analytic ik branches for kinematic cal fixes on 052 and 056
- fix cytometer experiment template
- fix 056 seed joints + check it’s using serial thing by jason
- service washer on 056
Interview Rutvik Interview strengths
- technically good weaknesses
- jumped to quickly into implementation
Overall, did a pretty good job - didn’t have to be prompted about how to deal with multiple cameras and how to thread to accommodate multiple client requests at the same time. Wanted to jump into implementation a bit too quickly, but talked through a reasonable design after prompted. Approach to AI code tools was also decent - reviewed generated code and identified several places that claude over-engineered. In the end, implementation didn’t work but his debugging was sensible systems debugging - and it seemed to be pretty clearly due to google meet holding control of the webcam. Tested a synthetic frame generation implementation which did work.
general interview notes: breaking things down into systems and building them piece by piece focusing on what is important walk through how they’re going to implement