How To Be An Outstanding CS Grad Student

A Few General Remarks:

• Keep your ego in check. Avoid trying to show off or brag. Yes, you made it to grad school. But so did all those other students. Now, the journey starts again — you need to make a transition from being the senior-most undergrad to the junior-most grad student. When you leave grad school, you will make another transition from being the senior-most grad student to the junior-most employee at the place you choose to work.
  
  
• Make a careful choice of career, and select Masters or Ph.D. accordingly; if you don't intend to do research, a Ph.D. is not the best use of your time. When you write a computer program do you think about other ways it could be written and wish you had the time to compare them? Do you routinely undertake fun projects outside of class that require many months to complete? Do you enjoy making a plan, creating a design, and completing an implementation on your own? Self-motivation and the ability to complete independent projects are characteristics of a good researcher.
  
  
• Enjoy grad school. Cover as many topics, and soak up as much as you can. You never know what will pay off later in your career.
  
  

Masters:

• Think and act as if you are a professional who designs and builds products (what industry calls an “engineer”). Seek successful faculty who have the most contact with industry and work with them.
  
  
• Consider each course carefully, and ask what you can take away from the course that may be of potential benefit later. Even if you do not intend to create computer programs or systems related to the topic at hand, ask what you can learn that applies to other areas of Computer Science.
  
  
• Strive to be among the top five students in each class, even if you think the material is not pertinent to your chosen career; don't convince yourself that you can afford to blow off a course just because you find it uninteresting or irrelevant.
  
  
• Whenever possible, implement the items being discussed in class, even when implementation is not a required part of the course. (The only way to develop your skills is practice; the only way to understand a topic completely is to gain first-hand knowledge.)
  
  
• Choose a project that makes you stand out from the crowd. Seek out the top Masters students, especially those who have had some industrial experience. Form a group and build projects outside of class. Find a professor who will give you 590 credit or do it on your own, but take it seriously. Format the code nicely, and write documentation. Yes, write documentation and get others to read it and use your programs. Remember that the goal is not to find code on the Internet but to figure out how to build things yourself from scratch. Discuss your designs and the techniques/tools you use with the group. Make sure your project is nontrivial (several thousands of lines of code) and a team project where you are in charge of dividing up the effort and coordinating.
  
  
• Develop writing and presentation skills. In industry, a project may be selected merely because the presentation is more polished than other presentations. Writing and presenting can be difficult for engineers, but stick to it. Remember that you do not have to compete with professional writers (e.g., English majors). Instead, you just need to be more competent than other engineers.
  
  
• Connect to industry. Skim the “trade rags” or on-line industry blogs in your area. Don't spend any real time — the objective is merely to get an idea of what's going on. Talk to corporate visitors, especially engineers. Take note of the lingo and current items of interest. Remember that a given corporate visitor may only have exposure to a small piece of one company, and may not have a broad perspective. Talk to several before making career plans.
  
  
• At all times, keep in mind: the more you learn now, the easier it will be to land a job and to perform the work required. A degree is not a hurdle to get over or a stamp that you collect on your way to a real job — think of the work toward your Masters as a plateau that you must reach and maintain, not only the knowledge but the ability to assimilate and learn new technologies and new skills quickly.
  
  

Ph.D.:

• Although you may hear people assert that a Ph.D. is “completely different than a Masters”, top Ph.D.s are not merely paper-mills that publish large numbers of thin papers — the best can do everything a Masters graduate can do plus more. So, read the description of Masters above, and plan to include programming projects in your student career.
  
  
• Think and act as if you are a professional researcher (either an industrial researcher or a faculty member).
  
  
• Take courses seriously — they are not merely a hurdle to overcome. Instead, assume each teaches you the background and essentials that you will need to do research. In each course, even in those that are unrelated to your chosen area, ask what you can take away that may help you in the future. Be aware that research doesn't follow strict boundaries — something you learn in the class may be useful later in your career. You'll be surprised to see that techniques from area X can be applied to area Y. Ask yourself how researchers initially developed the material in the course.
  
  
• Strive to be the best student in each class and impress each faculty member, even if the subject is unrelated to your chosen area. Strive especially hard to impress the top faculty members. As soon as an assignment is given, solve the problem. Then wait a day, reread what you have done, and polish the result. For a mathematical proof, clean up any ambiguities and make it more elegant. For a program, edit the comments and writeup to make them clear, and make the code as clean as possible. Don't ever give up, even if the assignment seems impossible. If you can't solve the problem by the deadline, dig in and solve it anyway for your own edification. Remember, the more you can do without searching the Internet, the stronger you will become.
  
  
• Seek out the top Ph.D. students, especially those who know more than you. Talk to them frequently about areas of CS and new ideas. Ask yourself what skills or knowledge they each have that you do not. Once you identify ways in which they are more skilled, work to improve yourself. Also ask what skills you have that they do not — to build a career, it will be important to identify your strengths and learn to exploit them. Look for interesting intellectual problems the group can discuss. Attend research talks and then discuss the topic, what you learned in the talk, and how the work can be applied/extended/improved. When you hear an idea in class that seems deep, shocking, or unexpected, write it down and discuss it in the group. Be intellectually competitive — try to find holes in other students' arguments.
  
  
• Practice thinking quickly. Learn how to analyze facts and how to form logical arguments in real time. Develop the skill of carrying on a technical conversation without hesitation and without introducing irrelevancies.
  
  
• In each task you do — whether it is research, course work, or even a mundane task — look for the fundamental abstractions. Ask yourself, how to describe the problem, potential solutions, lessons, and results in an abstract manner. Learn to identify the essence rather than the details; learn to characterize and assess.
  
  
• Practice technical writing. Although some researchers get by with ambiguity and obfuscation, the best strive to make their ideas accessible and understandable. Never assume that muddled prose means that a subject is intellectually deep — it usually means that the author cannot write well. At all times, strive to create prose that is crisp, terse, and accurate.
  
  
• Don't spend time trying to emulate someone else, even your advisor. Watch, learn, and adopt the skills you admire most, but do not merely emulate. Instead, find your individual strengths and build on them. Edsger Dijkstra summed up the idea: do only what only you can do.
  
  
• On choosing a topic: the choice of an advisor is more important than the choice of a topic. Choose an advisor whose career matches your inspirations, but remember that your career will differ from your advisor's in significant ways. Avoid choosing a topic merely because it is “hot”: you will be competing with many other grad students. Also remember that a Ph.D. dissertation is only practice — no one expects you to resolve one of the major open questions in computing. Finishing is more important than wasting years on a problem that you cannot solve.
  
  
• On publications: Albert Einstein once said, “I have little patience with scientists who take a board of wood, look for its thinnest part, and drill a great number of holes where the drilling is easy.” You cannot build an impressive record by accumulating a large number of meaningless papers. If you are aiming for the top, remember that one significant paper has more impact and earns more recognition than a hundred trivial papers.
  
  
• On choosing academic or industrial research: there is no optimum. However, your happiness may depend on the choice. If you are self-motivated, crave freedom, and enjoy teaching, academia may be best for you. If you like direction and want to see your work have commercial impact, industry may be best for you.
  
  
• On the relationship between research and teaching: if you want to learn a subject well, teach it. Undergraduates demand a teacher with perspective on the subject and the field, beginning graduate students demand a teacher who has mastered the subject, and Ph.D. students demand a teacher who can go beyond the facts and explain alternatives, tradeoffs, and consequences. Finally, if you want to determine whether a topic or approach is practical, teach a group of engineers who work in industry. They can easily sense whether something is practical or merely an academic exercise.
  
  
• Go beyond courses and beyond a single research project. Define a small, fun project on your own. It doesn't have to be “publishable” in the usual sense, but it must be entirely yours. Identify the question, devise an experiment, and carry out the investigation without guidance. Get other top Ph.D. students involved, but be sure you have at least one project that you create and control. After all, that's what a research career entails.