In my time at Michigan, I have had the opportunity to teach an array of courses, some for multiple terms. In 2012, I was recognized by the EECS department and won the 2011-2012 CSE Undergraduate Instructor Award.
In Winter 2016, I built a new course from scratch: Computing for Computing Scientists.
Towards the end of my undergraduate teaching career I observed a gap in the skills and knowledge of our students. A remarkable number of students in senior-level courses did not use build systems, version control, debuggers, or other common software development practices. While they may have been exposed to these tools, learning them amongst all the other coursework was lost, resulting in an achievement gap on projects that was wholly orthogonal to the comprehension of course material. With a groundswell of support from undergraduate students who reviewed the proposed syllabus, I created Computing for Computer Scientists (C4CS) to help address this, and taught it for the first time in W'16 to over 300 students.
C4CS is a 1-credit course aimed at improving the efficacy of software engineers. While the course centers around tools, it is not a series of tutorials. Rather we seek to develop an understanding for why a tool exists and how you might build such a thing, to build an intuition and lasting comprehension. One element students appreciate is that the course itself is very open, all course materials and course development are open source and available for anyone to explore at c4cs.github.io and github/c4cs. I have received the most positive feedback however, about the lecture structure, which is a mix of small amounts of traditional lecturing and then interactive sessions, me typing on screen with students following along, exploring how things work, and—critically—making, correcting, and learning from mistakes in real time.
One of the most powerful things the course does is attempt to eliminate the super-hacker archetype. We show that command line wizardry is not purely for the elites, it is simply a different language that has not yet been learned (and is unfortunately a bit arcane). During the interactive elements student questions consistently result in "live Googling", because even an "expert" doesn't know everything. We also aim to eliminate the notion that there is a "right" way to write code; that an elite emacs user is a fundamentally better programmer than someone using Sublime. The result of this effort is an increase in the comfort and confidence of students who had little to no programming experience before college. Numerous students, many from underrepresented minorities in CS, have sent private messages to the course staff thanking us for the course as they had been thinking of leaving CS as they were "not smart enough" for it, when really that was not the case at all.
EECS 470 is a Major Design Experience (MDE, the Michigan term for an engineering capstone course). It has a well-earned reptuation as one of the toughest MDEs at Michigan and was a very different teaching experience. As I was only taking one course this semester, I took on double hours supporting this course. We decided to change the course format and transform early-semester recitation sessions into labs. The course GSI, Pulkit Gupta, and I developed the labs that are still in use (as of W'14) today.
EECS 373: Design of Microprocessor-Based Systems (W'11, F'11)
I served as a lab instructor, a very different experience from 482. As a lab instructor, I spent no time lecturing. Rather, my time was spent helping to devlop and refine the labs and with a lot of hands-on time with students helping them through the tasks and challenges presented by the labs.
If F'11, in addition to lab responsibilities, I built a novel simulator project for the students. The project required students to build an ARM Cortex-M3 simulator, however, instead of the more traditional model of every student implementing a carbon-copy project, the entire class had to work together on one shared repository. A novel simulator architecture facilitated collaborative work. Students "registered" opcode masks with callback functions that "taught" the simulator core how to execute instructions (that is, the decode stage would scan all registered handlers and call the student's implementation of ADD, SUB, etc). In addition, students had to build (simulated) reset hardware and a software library to run on the simulated chip. The project was extremely interesting, and students reported it to be "unlike anything they had ever seen before". Students had to learn the challenges of developing and debugging their code, when the code of other engineers on their project is also new and unreliable. In addition, they had to deal with the challenges of co-operatively developing software and hardware. In the end, we found the project to be a much more accurate microcosm of real-world engineering - in particular new hardware development - and in some cases exposing critical gaps in the skill set of our engineering students.
- Those interested in more information are encouraged to download a copy of the project spec
Continuing the trend of building a class students had "never seen before", for the final exam (take-home), I built a reverse-engineering challenge. The exam should still be live, and those interested in trying it out can use the UMID 00011000 to try it yourself!
- Those interested in more information are encouraged to attempt the question themselves
This was my first teaching experience. At Michigan, undergraduate students are offered the opportunity to be Instructional Aides (IA's). An IA often covers discussion (recitation) sections for larger courses.
I was offered the opportunity to teach either 281 (Intro to Algorithms) or 482 (Operating Systems). I have always had a passion for low-level internals, and the opportunity to teach our OS course aligned best with my interests.
Over the four semester that I taught this course I worked with Peter Chen during the fall terms, and Jason Flinn during the winter terms. It was interesting to see different perspectives on what is, nominally, the same course.
As an IA, I led a weekly discussion session covering course material and held office hours to assist with comprehension of material and challenges with course projects. In addition to the prescribed duties, I held exam review sessions for the midterms and finals. This course is fairly refined and established (in fact, it's a bit of an institution at Michigan), and I am proud to be a part of the course's legacy.