How I started the journey? - Part I.1

It wasn’t how I thought it was, but it was how I did it.

“Do you still remember why you wanted to do it?” After 5.5 years, when I finally hung that piece of nicely framed wall decoration, I still remembered why I chose to do it.




It wouldn’t surprise me, if, in a few years, your memory becomes a bit blurry, and you couldn’t remember exactly why you started the journey.

Studying for a Ph.D. is an extremely demanding job, filled with moments of frustration and anxiety. Yet, it never fails to offer surprises and uncertainties.

You could start yours young and ambitious, determined to be a scientist, and ended up having a regular engineer job somewhere in industry; or, your goal was unclear when started, but your mind somehow drifted towards fighting an uphill battle of getting into some research institutes.


😎 There is one thing that is almost surely going to happen: there will be uncountably many moments you find yourself being very emotional: disappointment, happiness, fulfillment, etc. You will ride the ups and downs, smiling, laughing, or crying. It was quite an experience to me!

🤔 To most people, including your full-time peers, it would be quite a luxury to spend their early 30s in such a unique way. It was still surreal to me even after a few months into the program. I had a fantasy that I was living in someone else’s life, very peculiar.



I graduated from Lehigh University in 2022 with a Ph.D. in Industrial Engineering.

It is funny that my research had little to do with IE. I spent most of my research efforts cracking a specific subset of algorithms in the intersection of Computer Science and Optimization. A majority of the research groups in my field are under CS departments. If you know what ADAM or SGD is, you know 90% of what my research was about.

In particular, I was mostly concerned with the efficiency of training an ML model, and how to make the training autonomous. This is the field of designing adaptive optimization algorithms. It is still a cutting edge field, very little “known” and a lot of “unknown”. But, one day, we might be able to train a model at the scale of GPT4 in just a few (hundreds of) iterations from the scratch.

Alt Text
This was my office in Day 1. No windows, a bit dark, and 6 people sharing one room. Prior to COVID, I went to Lehigh campus and worked there every day for 3 straight years. It was where I had a lot of good memories and made a few really good friends.


You need to be a little bit contrary to start a part-time Ph.D. journey.

It might be the case that many people have thought or will think about it at some point of time in their life. But, I am pretty sure only few would actually do it.

Looking around the people I knew, friends, family or acquaintances, I could count, on one hand, those who actually started their part-time Ph.D. studies while working on full-time jobs.

🤔 The chance for you to know anyone, who had a full-time job and worked on a part-time Ph.D., is very small, if not zero. My experience could lead to a positivity and confirmation bias: by the time I knew a few, I already started mine. They somehow connected with me because they wanted to know they were not alone.

Looking back, I think I was a particular type of person, who never followed the conventional “wisdom”. Or, I should put it in this way: I never felt like following it even though sometimes I had to. From time to time, I often found myself having weird or even crazy ideas. I was certainly a contrary when I embraced the “torture”. Who would do a Ph.D. at an age of 30? Pretty sure that’s not a societal norm.


📖 Contrary, in some Native American cultures, is a person who adopts behavior deliberately the opposite of other tribal members. (source: Wikipedia)

Being a contrary momentarily is very easy. We all had some flashes of memories of believing the rest of world was wrong. Being a persistent contrary is not easy, and being a consciously persistent one is often the most difficult.


😮‍💨 Among those I knew who actually did it, over half of them ended up quitting theirs. A bit sad I’d say. But things like that always happened, and it will happen to some of you. If I would quit at some point, the best thing I could have done was to quit before it ever started! Then, you would not be able to read my stories.

We are living in a global tribe of 8 billion population, or over 100 billion including those dead for good. If only a minority group of tribe members is blessed, my chance of wring my own success story would be much bigger than the average if I was a contrary. In my view, doing whatever every one does or think I should be doing is fundamentally against my odds of living a successful life.


😎 If you are in early 30s, you are far from being too old to start a Ph.D. program. Before I got accepted by Lehigh, they had a guy at 70s retired from his surgeon career.

That said, it is still too early to call whether or not it was the best way to start my 30s, let alone justifying if I was really a wise guy. But, at the time of making that decision, I knew I was very conscious and would take a full responsibility of any consequences down the road.


🤔 It seems a riskier job to work on getting a Ph.D. in early 30s than, say, in early 20s. But, I would argue that it all boils down to the risk-reward calculation. If I had a 7-figure salary job, I probably wouldn’t want to be distracted by a doctoral study. But, if I can handle a 6-figure salary job while doing what I loved doing, it probably wouldn’t be that bad. Taking a calculated risk and knowing the opportunity cost is the way how I looked at it.

In case I would ever regret doing it, I actually tried very hard to think of any validate reasons of not doing it. At the end of the day, I couldn’t figure out anything convincing. So, I knew I couldn’t get away from the idea, and I should just follow my heart.

I still hold my belief firmly till this day: following through a Ph.D. program is the easiest way to get knowledge and, even better, to create knowledge. It is a highly productive process, which has been validated by its own history. Every piece in the process has been repeated millions of times in the history, and the chance of having any unique situation is very low. Although the system does not guarantee any individual success, I had no doubt in myself.

😎 If you don’t have to pay the tuition out-of-pocket, working on a Ph.D. is really neat! You get to acquire all kinds of skills without paying a dime! Compared to those 9-5 labors, you can take a break and stop worrying about catching up with trending skills. It was a great deal to me!

Enough for justifying being a contrary! Most of time, I am not against the societal norm. But, I had my own belief of what I should be doing. I guess that’s also why I managed to finish the program.

Zheng Shi
Zheng Shi
Staff Machine Learning Engineer | Ph.D., Machine Learning & Optimization

Deep Learning, Machine Learning, Optimization Algorithms, and Data Science.

Next
Previous