Sei sulla pagina 1di 38

Career advice for PhD students: How to get the most out of your time in the PhD program

Preamble

Why am I doing this?


Not many resources to learn how to be a successful PhD student trying to help you

Faculty create new knowledge and next generation of researchers

A professor is as good as his best student

Why now?
As every September, we got fresh PhD students
I might soon forget my PhD student experiences

Talk applies to any CS PhD student despite influence from personal experiences and systems/networking background Acknowledgment: I admit to stealing advices from many successful people (too many to be listed)
2

Outline

PhD student stages


Thinking about doing a PhD Taking classes and getting involved in some research Choosing research area, topic, and advisor Doing research Writing the thesis Getting a job

Slightly different view of these stages


1. Student: I know everything; Advisor smiles 2. Student: I dont know anything; Advisor: Lets talk 3. Advisor: Lets do X; Student: Youre wrong because of Y and Z
3

Why are you getting a PhD?

Prerequisite to a research career


A PhD degree should ensure that the student can later take on independent, long-term research commitments

The work required to earn a PhD is not worth the effort if you dont intend to do research
You can do better with an MS degree in such a case

How do you know if research is for you?


Have inquisitive mind and critical thinking Like to understand how things work Like to identify problems and come up with solutions Did some research during undergraduate studies and liked it More philosophical reasons: dream of changing the world, good way to have a legacy beyond your family
4

Bad reasons for pursuing a PhD

Afraid of going out in the real world


If you never had a job and not sure about going for a PhD, go and work one-two years

Ego Impress your girlfriend/boyfriend/parents Opportunity to work/emigrate in US


OK if your goal is to do research in (still) the best place for that in the world Otherwise, working very hard for something that you dont care much while living on a PhD stipend will soon make you unhappy

Money (i.e., amount of money you make is more important than what you do)
While starting salaries of CS PhD graduates are good, can reach higher salary if you worked since you got your BS/MS degree

Plus money earned during that time

What qualities do you need to be successful in the PhD program?

Passion and Self-Motivation


Doing a PhD is a life changing decision

Be sure that this is the path you want to follow in life (yes, its normal to have doubts sometimes)

Perseverance and Self-Confidence


It could be heartbreaking to work hard for one-two years and get your paper rejected Trust yourself (and your ideas) and dont give up

Independence
Its your PhD; you should know what you want to do, how you want to do it, etc.

Obviously, you need intelligence


Many times you dont know how smart you are until somebody challenges you
6

CS department expectations*

Take qualifying exams after first year and pass them all after second year
Proves that you are good enough to continue in the program

Find advisor and choose thesis topic after second year Defend thesis proposal by the end of third year
Not very strict deadline (depends on progress and advisor)

Defend thesis by the end of fourth year


Can stay longer if necessary if advisor awards you RAship

Take a number of courses and maintain a decent GPA (e.g., 3.5) throughout these years

* refer to full time, department-supported students


7

Advisor expectations

Every PhD student must have thesis/research advisor Advisor decides when student is ready to graduate
Process very similar to apprenticeship Thesis committee makes sure advisors decision is correct and gives feedback to improve work

Each advisor has own requirements, but they can be generalized as:
Have enough background in CS and depth in your research area Work on one or multiple projects and publish the results in several good conference/journal papers

Be able to clearly present your ideas and results


Write a good thesis

Your papers and thesis must include your novel ideas


Of course, they include your advisors ideas as well
8

First year

Get involved in research!


Ask professors with research interests matching yours Combine reading with working on a small part of a project Steal tricks of the trade from advisor and more senior students

Classes and the qualifying exam are required, but dont spend more time than necessary on them
Nobody cares about the grades of someone with a PhD degree

Dont get bogged down with teaching/grading


Need to do a decent job, but make sure you dont work more than the required 20 hours/week (many times you can work a lot less)
9

TAship vs. RAship

RAship is better
Can spend time on you research instead of teaching Being awarded an RAship means youre doing well Since RAship comes from a grant, the advisor will ask you to work on the project defined by that grant Advisor can ask you to work on demos or robust implementations as required by grant (which are not necessarily research)

TAship has some advantages as well


Independent to work with several professors before deciding about advisor

Teaching experience required if you think of academic career


Teaching helps you improve communication skills Every PhD student should teach at least one semester
10

Choosing research area

Dont celebrate too much passing the qualifying exams


You are expected to pass

Choose area based on your research interests


Must like it; otherwise, the next few years will be painful Dont choose it just because you can get an RAship

Need to think strategically as well


Is this a hot area? Will you get a good job in this area after graduation?

Hard to predict if certain areas that are hot now will still be hot in 4 years

11

Choosing advisor

Should be compatible with advisor/get well together Tenured advisors


Have more experience, could have more money, could have more connections Dont push you hard, dont have time to work closely with you

Tenure-track advisors
Will push you hard (their future career depends on your results), but will work with you (i.e., co-authors of thesis) Might have more up-to-date information about job searching

12

Choosing thesis topic


Its your topic, but the advisor must approve it Its rare to know the topic from the moment you start working with advisor
If work supported by a grant, the general topic is somewhat clearer

More common to work on several related topics in your chosen area


First ideas might not work, new ideas could come up Some will be more successful than others publication-wise Many times, thesis will define a common framework for topics covered by publications

13

Take ownership of your PhD

No one is responsible for getting your degree but you


Faculty set up opportunity, but its up to you to leverage it

14

Doing research (1)

Be proactive!
Dont wait for advisor to push you

Reading papers
Develop critical thinking: identify both strong and weak points Advisor will point you to important papers as well as conferences and journals in your area You responsibility to find more papers starting from these pointers Must read a few papers every week Read outside your area as well Follow technology news to know where the world is going

Let advisor/colleagues know about interesting things you read


Robin Kravetss advices for reading/presenting papers

http://www.cs.njit.edu/~borcea/reading-papers-talk.pdf
15

Doing research (2)

Identifying important and hard problems


Learn to differentiate between cool problems and junk

Advisor will offer a lot of guidance

By graduation time, acquire good taste for selecting problems

Problem solving/design
Always ask yourself: whats the novelty of my solution?

Also: how is it different from/similar to alternative solutions?

Advisor suggests a potential solution


Never go back and say doesnt work!


Instead, say X didnt work, but how about Y or Z?

Dont get upset/discouraged if advisor points out drawbacks in your solutions its technical, not personal 16

Doing research (3)

Implementation
Except for purely theoretical CS, will have to implement your ideas Every successful project goes through this unglamorous, hard phase Design is more fun than implementing it No magic here: work hard! Dont suffer in silence if you dont know how to implement something or have troubles with a bug ask colleagues or advisor for help

Evaluation
Prove that your solution works as claimed Should know from the design time experiments and metrics Form a hypothesis: what type of results you expect Experiments contradict hypothesis: think of potential reasons and discuss them with advisor

Work in the lab a significant amount of time


Learn from interactions with colleagues/advisor
17

Mutual trust between student and advisor

Trust advisor and earn his/her trust (e.g., through good work, reliability)
Advisors, being human, are not perfect, but try their best to help

Almost everyone goes through periods when doubts advisor (the converse holds as well)
Papers getting rejected Different opinions on how to proceed with a project Seemingly advisor cares only about his career

During these periods, remember the advisor/student symbiosis


Advisors work hard to get grants to support your work
You work hard to produce results that will enable new grants Typically, what is good for advisor is good for student, and what is good for student is good for advisor
18

Communicating your results

Clear communication separates top students from average


An unknown brilliant result is useless

Write and publish papers in conferences/journals


If you didnt write it down, it didnt happen Publish or perish Reviewed by peers Hard to get accepted (good publication venues have 10-15% acceptance ratio) Can start small with conference posters or workshop papers

Talks
Presentations of accepted conference papers (or invited talks) Good chance to convince people that you did great research

Successful researchers spend 50% of time writing papers and preparing talks
19

Writing papers

A lot harder than you think!


Good results are not published due to sloppy writing

Ask advisor for models of good papers Get feedback from advisor early and often; then re-write Read Shrunk and White book on writing One idea per paragraph
Do paragraphs follow one another in a logical structure?

Typical structure: abstract, introduction, related work, design, implementation, evaluation, conclusions Have clear abstract/introduction
If vague or poorly written, reviewers will just look for reasons to reject afterwards

Dont claim more than you did


Distinguish between will do and have been done
20

Conference talks

Goal is to make audience read your paper and talk with you
Emphasize the main idea, skip some details Shouldnt follow too closely the structure of the paper Pay special attention to motivation

The more illustrations, the better


A picture is worth 1000 words Dont take this talk as model

The more you practice, the fewer surprises during the actual talk
Time management is your responsibility; be prepared to skip slides

Show excitement
If you are not excited, then why would anyone else be?

Be clear, firm, and polite when answering questions


Show belief in your work
21

Attending conferences

Typically, you go when have an accepted paper


Could ask advisor to pay or get travel grants to go to top conferences even if you dont have paper there

Check technical program ahead of time and identify papers/people of interest


Goal is to do networking, not just hear technical talks
Take advantage of coffee breaks/lunches/receptions to talk with people Be prepared to initiate conversations and introduce your work (prepare an elevator pitch) Get contact information from people you want to stay in touch Learn how top researchers present their work and answer questions

People you meet there can hire you, review your papers, or become future collaborators
22

Summer internships

You should go once or twice


Get real-world experience, make connections Must do it if plan to work in research labs/industry Go in research oriented places

Doing an internship just for money is not worth the time

Decide together with advisor when and where to go


Advisor can help you go to good places (e.g., IBM Research, Microsoft Research)
Better go once you have at least one publication; can select internship that allows you to work on related topics

Be aware that they can delay graduation as summers can be very productive research-wise
Cant have the cake and eat it too
23

How much should you work?

Work only the number of hours you are paid!

Dont let the master class exploit the workers!

Students in high-ranked schools work between 60 and 80 hours per week


Faculty spend a similar amount of time Dont get fooled that you do better than some colleagues while spending a lot less time You will compete for jobs with students form other schools as well

Citing my advisor: school breaks are for undergrad students


Good time to work in case you have teaching duties The advisor has more free time to help you
24

Dont have time to finish all your tasks?


Must acquire time management skills Write down your tasks (both work-related and personal), set deadlines, and categorize them function of importance Randy Pauschs graph for task time management: Importance Continue with these tasks

Obviously, finish these tasks first

Urgency
25

More on time management

Dont have time for personal life?


Some personal tasks must have high importance Family/friends help you avoid going nuts According to previous slide, you might end up not doing urgent, but not important tasks; its ok, the world goes on

Know yourself and manage advisors expectations


Learn to estimate accurately the time it takes to do certain tasks Learn to say no if its not possible to do a task before a deadline Try hard to respect deadlines once you agreed to them Inform your advisor as soon as you are getting behind the schedule

26

When to graduate?

Graduating as fast as possible might not be the best idea


This is not the Olympics where the best finishes first Should become a well-rounded researcher, not just someone very narrow expertise Working on larger/higher impact project might take longer, but help you become a better researcher and get a better job Taking classes outside your area and attending seminars/talks can improve your overall background Doing paper reviews or helping advisor with grant proposals can take time, but are invaluable learning experiences Job market conditions may delay graduation

Taking longer than 6 years not good either


Potential employers dont like it Even advisor might lose interest in you
27

Thesis (1)

Thesis: one sentence to describe your contribution to the progress of humankind Dissertation: the 100s pages that prove the thesis

Dissertation is very much a collection of your publications


Of course, need to link them well under one clear thesis Also, need extensive related work and potentially more experiments

Thesis proposal
~= thesis without a chapter or two Not as important as you may think because early validation of your research comes from good publications

Form thesis committee and get feedback from committee members

Both student and advisor must agree on committee members

Contract between you and committee: agree on content to be added in the final thesis

28

Thesis (2)

Finish writing during your final year


In parallel with job searching Models: theses that received ACM awards

Thesis defense is reason to celebrate


Advisor/committee wont allow you to defend if not ready

Not a good idea to defend if you dont have a job (especially for foreign students who plan to stay in US)
Unless you dont receive support any longer

You could get job before thesis defense


Risk: you might never get the drive to finish

Useful things to know about PhD thesis research by H.T. Kung


http://www.eecs.harvard.edu/~htk/thesis.htm
29

Job searching

Once advisor confirms you will be ready to graduate that year, prepare:
CV (long, not the typical 2-page resume)

Research statement (at least 2 pages) outlining your research contributions and future plans
Teaching statement (if applying to academia) outlining your teaching experience, teaching philosophy, etc List of references Have them ready by early December

Most academia and research jobs are posted by January


Must submit the above-mentioned documents by their deadlines

Have your job talk ready by January Learn about research interviews by January Wait for call/email and hope
30

Job in academia

Research universities have similar starting salary with research labs (but doesnt increase at the same rate)
Teaching university have significantly lower salary (and no research)

Flexibility to choose research topics


Can work on fundamental research and explore higher risk ideas Need to get them funded through grants

Can publish and go to conferences more often than in research labs Can make your own schedule
In the beginning, you work more than in industry

Can influence people directly through education Safer job (after tenure)
31

Job in research lab

Over a number of years, salary will be slightly higher than academia (could go for management positions as well) Can have impact on real world through products incorporating your ideas Research topics need to be in line with companys goals and approved by managers
Short-term profit-oriented research may preclude you from working on fundamental or high risk topics Working in an R&D department is even more about practical research that can quickly turn into profit Still need to worry about funding (convince your managers to invest in your ideas)

Cant publish everything


Patents first, publication later (if at all)

Job safety depends on company health & market

32

What do interviewers look for in your CV?

Thesis title, research interests, and name of advisor


The advisors reputation matters a lot

Research contributions
Projects you worked on and their main results Software distributions

List of papers & talks (& patents if any) Teaching experience (for academia) List of references
Reference letters are very important

CS community service (e.g., conference/journal reviewer) NO!


GPA Programming languages, tools, etc (you have a PhD in CS! Youre supposed to either know or be able to learn everything)
33

Job talk

Single most important part of your interview Two main purposes


Sell yourself Sell your research

Write down 3-4 ideas youre going to say per slide


Practice and remember those ideas

Do dry runs with advisor, colleagues, friends Videotape yourself and try to improve after the shock of watching the recording has passed Practice questions and answers More information on job talks and interviews from Jeanette Wing
http://www-2.cs.cmu.edu/afs/cs/usr/wing/www/tips.pdf
34

One-to-one interviews

Typically, 30 minutes about your research and everything else They look for
Creativity Brainpower Independence Technical skills Leadership Energy Fitting in

Be prepared, articulated, honest, genuinely curious


Ask questions about the persons research Ask questions about the place to see if its right for you OK to engage in less technical discussions (e.g., benefits, housing)
35

Selecting a job

Congratulations, you got several job offers! Many factors to consider besides money Reputation of the place Can you grow there? Possibilities for promotion? Will you get along well with your colleagues/bosses? Geography Two-body problem Cost of living Quality of schools Are you a city person or more of the outdoor-type?

36

More readings instead of conclusion

How to Be a Good Graduate Student by Marie desJardins


http://www.cs.indiana.edu/how.2b/how.2b.html

So long, and thanks for the Ph.D.! by Ronald T. Azuma


http://www.cs.unc.edu/~azuma/hitch4.html

You and your research by Richard Hamming


http://www.cs.virginia.edu/~robins/YouAndYourResearch.html

Technology and courage by Ivan Sutherland


http://research.sun.com/techrep/Perspectives/smli_ps-1.pdf

How to have a bad career in academia by David Patterson


http://www.cs.berkeley.edu/~pattrsn/talks/BadCareer.ppt

Paper writing and presentation by Armando Fox


http://www.cs.berkeley.edu/~fox/paper_writing.html
37

Your time in the PhD program is a unique experience: Enjoy it! Good luck and make us proud!

38

Potrebbero piacerti anche