Aditya Gupta - June 30 2018

My thoughts and experience so far teaching at Stanford

I know that the title of this article may seem as though I am a professor of some sort at Stanford, but I’m really teaching a machine learning course at a tech summer camp hosted on the campus. I have been working on Stanford’s beautiful campus for over about a week now, and wanted to leave some thoughts on what it’s been like to work at one of the most prestigious universities in the world.

When I got the opportunity to teach machine learning, one of the many subjects I’m passionate about, I didn’t hesitate to jump on it.

Many of my peers are taking the conventional route which involves beginning in an internship of some sort at some decently known company with the occasional friend or two at Google or Microsoft. I didn’t want to do that. This is not to suggest that this route is below me or that I am not cut out for getting direct industry experience. Something about teaching just sparks a passion in me that most things can’t. There’s a certain sense of excitement, joy, intensity, and nerve all combined together in a kind of emotional smoothie that comes with teaching. The risk of embarrassing yourself in front of all your students is high, but the reward of delivering an immersive and memorable experience is higher.

I’ll run through all five days of instruction from my first week, explaining the experiences I gained from teaching at Stanford.

DAY 1:

In all honesty, I didn’t really know what to expect for day one. This being my first time teaching, it could have gone terribly. Luckily enough, the day went well. I found myself engaging my students in a review of basic Python concepts, having them come up to a whiteboard and solve problems in front of their peers then explain their intuition behind those problems. Something I found to be particularly helpful and fun for the students was pulling simple practice problems from the likes of leetcode and codingbat, and modifying the structure of the problems to let the students think a little bit more than they normally would. Of course, this being a machine learning course, I got a lot of questions involving the statement, “when will we be doing the actual machine learning stuff?” I figured that I’d play by ear, but the answer was simple.

Soon.

DAY 2:

Day two was interesting. Not only was the planned schedule hectic and conflicting with my lesson plan for the day, but we also had an unexpected guest lecturer arrive to talk to us about the advent of cyber security as a fundamental tool for the DHS (Department of Homeland Security). Princess Young, the speaker that came to talk to us, was a phenomenal individual. She spoke about how cyber security plays a fundamental role in governmental affairs, and the entire time, my students were loving it. After, we headed back to our lab where I planned to lecture about the importance of neural networks in machine learning. Of course, since the guest speaker was a bit of an unexpected addition to the schedule, my lecture time was essentially eradicated, so I would need to push my lecture on the tenets of basic machine learning principles back to day three.

Fair enough.

DAY 3:

Now I was getting deep into the week and really needed to progress in my instruction. My colleague Lukah, who was also a machine learning instructor, had suggested we take our students outside of our lab and into the lounge, to allow for a more relaxed and casual learning experience. Great idea Lukah. We proceeded to combine our classes and lectured about the structures and applications of neural networks, the architecture of computers and bit allocation, and how we could apply these concepts to build our final project which would be an image recognizer.

It was soon after this lecture that I realized I probably was not instructing at a pace satisfactory enough to cover all the content advertised in the course. Or was it rather that I was being given too much content to shove down the mouths of confused students forced to attend my class because their parents saw “machine learning” and salivated while pulling out their credit cards. When frustration set in because projects weren’t coming along as planned, it was certainly the latter. But in the end, I found a way to make it work despite feeling like I was perhaps giving too much for my students to learn. But then the question became, how would I answer my student’s parents when their kids didn’t know how to answer their questions about their final projects?

Shit.

DAY 4:

Now stuff was getting real. My goal for the day was to get at least half of my students up and running with their image recognizers, and to have those image recognizers work within 80% to 90% accuracy rates. The learning algorithm we were using was a relatively simple one, utilizing a convolutional neural network and the likes of linear regression and gradient descent. I proceeded to lecture again on why we were doing what we were doing. Now, things were starting to make sense for my students. My class had ten kids in it, and by mid day Thursday, three had already completed their projects and were adamant on moving on to something cooler or improving upon what they had developed with their image recognizers. I was pleasantly astounded at how naturally the intuition behind what I was teaching had come to them. And their programs weren’t just working, but they were working well. At times, they would correctly predict 10/10 images correctly, which was absolutely amazing. But then it set in. That was just three kids. What about the other seven? What were they going to do? Half of their projects weren’t working and I found myself literally running back forth around the lab trying to solve one issue after the next for each one of them respectively. There was seven of them with an unending flood of reshaping errors and typos that were hindering the progress of their entire program, and just one of me. I needed to figure out what to do before all of their parents arrived on Friday to see their final presentations and evaluate me as an instructor. Now, I knew that most of the kids liked me. Some of them would even willingly sit with me at lunch and continue to ask me questions about what I wanted to do for a living and what I did for fun, which was honestly an amazing feeling. I feel that day four was probably the most pivotal day of the entire week in that, while I knew the next day was going to be rough with presentations, I had the satisfaction of knowing my students were enjoying themselves.

But it’s the parents’ opinions that matter right? After all, they are the ones paying money for their kids to be in my class.

Shit. Again.

DAY 5:

The day had come. I had spent the entire previous afternoon trying to get as many final projects completed as I could. Lukah, my colleague whom I mentioned earlier, was dealing with the same issues. We were running out of ideas of how to make things work before the parents arrived. At this point, I am not sure if it was the sense of urgency and adrenaline running through me, luck, or both, but Lukah and I were able to get all but one of our students’ projects working. It was as if our pleas had been answered, because had things turned out differently, we probably would have received some very strongly worded emails from angry parents. The parents arrived for the showcase to see what their kids had been doing during the week in my class, and for the most part, seemed pleasantly satisfied. I got lots of questions asking, “what does this do” or, “how did my kid know how to do this.” In all honesty, I was so exhausted by this point of the week that I basically gave the most generic answers possible so I could get home to my well deserved nap sooner than later. This is not to say that I was disinterested by the curiosity of my students’ parents, but rather that I was so burnt out from a week of running around and managing one of the most difficult courses offered at one of the most prestigious universities in the world. Besides this, the presentations went well and I wrapped up my first week as a teacher.

Mission successful.

It is clear to see that teaching, no matter what the subject may be, is no simple task. At any time, anything can go wrong. But, the rewards of pushing through the difficulties are so satisfying, that quitting isn’t an option. It’s because of the students that I got to work with, and the positive attitudes that they maintained no matter how difficult the subject material got, that make the job of teaching as fun for me as it is. I will continue teaching at Stanford for the duration of the summer and look forward to further updating the experiences I gain.