Thoughts on Leaving
A collection of thoughts from different places I’ve moved on from
Relax, I’m not leaving my current job right now. I’ve been writing about many different aspects of my work experience, and I think graceful exits are just as important as graceful entrances. I’ve only quit one company, but I have left two schools and two positions internally at Apple. There is something to be said about a graceful exit whether it is voluntary or not. Here are some of my experiences and what I did to exit with as much grace as possible.
Don’t be afraid; change is guaranteed.
Leaving University of Detroit Mercy (UDM)
At UDM, I poured my heart into our senior design project, an autonomous vehicle. I re-wrote functions in Matlab to squeeze out every bit of efficiency, and the result was 3,000 lines of code running three methods of lane detection, merging the results in 43ms.
After competing in the Intelligent Ground Vehicle Competition (IGVC), I had two months before graduation to document my work. I wrote two papers, and I worked to pass off the code. It was a bit laborious because I hadn’t commented my code well, and my variable naming convention was the worst (it’s better now). Imagine you watched Pulp Fiction and decided to name all your functions based on lines from the movie.
I was starting to stress about going to graduate school with this documentation unfinished, and I asked my mentor about it. He told me to go to grad school and forget about us. Forget about UDM, about the vehicle, the documentation, leave it all behind. He said focus entirely on the work you will have in front of you. Words to live by.
Leave it all behind
Leaving Notre Dame
The effort put into senior design project paled to the effort required for my dissertation. I faced major challenges particularly little oversight. I was in control, and I had to manage my project and get it done. My advisor advised, but rarely was I ever managed. I wasn’t used to being responsible for everything with respect to my project. The irony of grad school is that undergrad is all about teaching team work, but dissertation research in computer vision usually is an individual sport.
It was easy to leave Notre Dame because I was sick of living on a graduate stipend. My salary in 2 months at my first job was more than my salary for the entire year as a grad student. I had taken my mentor’s advise, and I threw myself into my work without looking back, at least, not too much. My first job at DSC though was different because it was still in the field of 3D face recognition. I was excited to continue working in my field of expertise.
Be proactive
One thing I did do though was continue to try to publish my work. I didn’t end up with more publications, but two years after I left, I had a chance to get full control of my patentapplication for my dissertation. Three office actions and one year later, and I was granted my first patent.
Leaving Digital Signal Corp (DSC)
I like to say I left voluntarily, but the next two years showed that the company was going under, and my time was limited there. I was lucky enough to see that before it happened. However, I was searching for a job for three months before I left. I knew my departure would be imminent or else I was gearing up for some major depression.
As a point of personal pride though, I continued to give my all to my job. I didn’t want to slack off, and I saw it still as a job I was being paid to do. Since I was still being paid, I would still work as usual. I didn’t want to leave and people say good riddance. I struggled with the possibility that my being upset would affect my job or my character because in the long term, I could have picked up character defects as a result of disliking a job and having to stay there for so long.
Give two weeks
I was kind, and I gave two weeks after getting a job offer from Apple. I was ready to help pass off code and results. That didn’t work out as well as I thought because I don’t think anyone was prepared for me or anyone else to leave. They should have jumped on the opportunity to get all the knowledge transferred as quickly as possible so they didn’t lose anything.
One point of contention was a single page they wanted me to sign. I was asked to sign a page verifying I left all content that was theirs and returned all hardware. I told them I would leave everything, and I did, but I didn’t want to sign anything. I had no benefit of signing anything; they weren’t offering money. They were afraid I would screw them over, which I didn’t. I did leave all my documentation and code. I didn’t want any of that at my new job, and my new job didn’t want me to bring anything. I was worried they would accuse me of stealing something and sue me.
Don’t sign anything without benefit.
Leaving Apple Watch Health Algorithms Group
I was excited to work on Face ID, but leaving the Watch group was sad for me. I felt like I left three times because three times I switched roles within the group. First was passing off wrist detection algorithms to someone else, and while this was relatively easy, I was sad to say goodbye to my baby.
Then I had to pass off my work on Background Heart Rate (BGHR) to a new engineer as I was preparing to go to Face ID. I was held over for two months to work on a feasibility study of an unreleased feature. I then had to pass that off. In all three cases, I worked to make sure I gave the best understanding of my work as I could, and I made myself available for questions and discussion. I was also proactive in going to the new engineer taking over to talk to them and make sure their questions were answered in case they were shy to ask.
Plan from the beginning for a future hand-off
Semi-Leaving Face ID Algorithms
Initially, I started working on core Face ID algorithms. While in that group, I worked on Design of Experiment for user studies, failure analysis, and deep net training. I’m quite a curious fellow, and my director thought it would be a better fit if I started a group to find all the variables we didn’t know about. This task wasn’t quite covered by quality assurance because they dealt with mostly what we knew, and it also wasn’t covered by our user studies team because they ran larger studies and were not as agile to such a task.
There was a large list of unknowns because nobody had lived on a phone with Face ID. We had to imagine how people would use the phone and in what ways different from Touch ID to see what data to test. This involved designing the data collection, collecting the data, and analyzing it. We also did any engineering studies requested by others and analyzed the resulting data to understand if we needed more data.
Continue working hard before, during and after a transition
This new team was still part of the algorithms team because when using a deep net, the data is the algorithm, but we had a different kind of flexibility and oversight. I technically switched in December, but I was in the middle of large make-up study I had spent months designing and doing small engineering studies to help support the DOE.
Switching this project to someone else would require quite a bit of work, and around the time I was supposed to move, I found a lot of incorrect labels from the big user study I designed. I pushed to stay on it a few more weeks so I could make sure I fixed all the labels and finish the analysis. I also picked up another engineering study from my soon to be previous manager, and I worked hard to get it done on time. I wanted to do a good job regardless of how my role was changing. I did transition roles, and in this case, whenever people had questions, my office only moved down the hall.
Leaving Lessons Learning
- Don’t be afraid; change is guaranteed.
- Continue working hard before, during, and after a transition
- Leave it all behind.
- Don’t sign anything without benefit.
- Give two weeks of your presence.
- Handing off work:
- Do due diligence
- Be proactive
- Plan from the beginning for a future handoff
If you like, follow me on Twitter and YouTube where I post videos espresso shots on different machines and espresso related stuff. You can also find me on LinkedIn.
Abandon Ship: How a startup went under
