Ask Slashdot: Scientific Research Positions For Programmers? 237
An anonymous reader writes "I recently (within the past couple years) graduated from college with a bachelor's degree in Computer Science and currently work as a programmer for a large software consulting firm. However, I've become gradually disillusioned with the financial-obsession of the business world and would like to work for the overall betterment of humanity instead. With that in mind, I'm looking to shift my career more toward the scientific research side of things. My interest in computer science always stemmed more from a desire to use it toward a fascinating end — such as modeling or analyzing scientific data — than from a love of business or programming itself. My background is mostly Java, with some experience in C++ and a little C. I have worked extensively with software analyzing big data for clients. My sole research experience comes from developing data analysis software for a geologic research project for a group of grad students; I was a volunteer but have co-authorship on their paper, which is pending publication. Is it realistic to be looking for a position as a programmer at a research institution with my current skills and experiences? Do such jobs even exist for non-graduate students? I'm willing to go to grad school (probably for geology) if necessary. Grad school aside, what specific technologies should I learn in order to gain an edge? Although if I went back to school I'd focus on geology, I'm otherwise open to working as a programmer for any researchers in the natural sciences who will take me."
Dichotomy (Score:5, Insightful)
I think a solution a lot of people find is to split their day: they pay their bills with a job they can (just about) tolerate, and then use their free time to focus on their passion, perhaps in a small community (cf. FOSS development).
Also, academia is no paradise either: it's not so much about focusing on what you are interested in, but rather focusing on where there is funding, and where you can find your own niche. It's surprising and depressing how many niches are already filled: it's like trying to find an empty shell on the ocean floor.
Don't know how is it in the USA, but ... (Score:3, Insightful)
I can talk from my experience in Europe. Although you may have the experience and knowledge to do the research successfully; going to grad school will open many doors. You will have access to information about ongoing projects, publications, etc ... And by the way you will fill some possible weak points in your knowledge about the subject.
About technologies; you must be flexible; just know how to program, not on a specific language. Anyway, I recommend you to get to know (and learn to love) Matlab.
Re:yes, there are a reasonable number of positions (Score:5, Insightful)
One thing that should be clarified here: with these sort of programming roles there is no direct access to academia. In this way it's not like finance where, for instance, people can go from back office to front office if they show enough promise and interest: one does not simply go from research programmer to reseacher.
From the sounds of it, OP would be best served by going into academia via a graduate program.
R language (R is too short for a Slashdot Subject) (Score:5, Insightful)
No such thing (Score:3, Insightful)
Seriously though, I think, with the exception of the "Alex P. Keatons" among us, virtually all programmers would rather work doing some sort of pure research for the betterment of humanity, than helping some sycophantic management team please the board/stockholders for yet another quarter.
Reality of the situation, though, you (and I, and all of us) have chosen the very same thing you claim has disillusioned you. You have chosen to want a paycheck. Make no mistake, for every one software engineering job position you see posted, you can find a hundred good causes that need volunteer coders. Except, good luck getting a steady paycheck if you go that route - Short of actually becoming a professor, you very much need to treat it as an act of charity.
Which leaves you to ask yourself: Can you really afford to live without a paycheck? If you can't answer "yes" without hesitation, hey, they don't call it "work" because we go there to have eight hours of fun every day.
As a compromise solution many of us have taken, do your good deeds on the side. Get that paycheck, and put 10-20 hours a week into a FOSS project, or helping the local foodbank set up a useable LAN from their pile of 15 year old mostly-DOA donated junk, or if you still have a few "in"s at your university, ask a few of your favorite non-CS professors if they have any projects that could use your skills (almost all of them do). But make a living first and foremost.
From my experience (Score:2, Insightful)
I work at a large research organization. I'll tell you how it is here, it will be similar at other places:
* We have research staff and non research staff (lawyers, personal assistants, software engineers, ...) ...
* All research staff must have a PhD in the field of their research position. I.e. if you want to do research, do a PhD first.
* Software engineers don't need a PhD, but we require a bachelors in IT or equivalent experience.
* Software engineers assist in research, but do not lead it. I.e. you don't get to work on your ideas, but on somebody else's. Still, it's research and some of it can be argued to be for the good of mankind.
* Almost all research is not as exiting as it is cracked up to be. Direct connections to the good of mankind are very rare.
* Most research projects are very small and you may be the only software engineer on it. Not all software engineers work well in such an environment.
* Most software you produce is very alpha and never gets further (run once, point proven, let's move on). This can be frustrating and also bad for your CV since you can't really claim you shipped a product for real customers.
* Work is not different than interesting jobs at industry such as IBM, Microsoft, Google,
* These days the research world is very financially obsessed, and research projects are most of the time determined and restricted by what your group can get funding for (rather than what is for the common good).
Having worked at NASA for such folks (Score:5, Insightful)
I can tell you that with out a PhD, your are viewed as little more than a trained chimp. Masters in both CS and Applied Math seemed to mean nothing, the fact that these so called doctors were incapable of writing more than 4 lines of intelligible code was beside the point.
It was fairly annoying, and none of my work is cited in their papers.
Scientific Programming (Score:5, Insightful)
Three words: Math, math, and math.
If you don't have the advanced math skills, your use to a scientific research effort will be limited.
Re:yes, there are a reasonable number of positions (Score:4, Insightful)
I actually moved in the opposite direction from a pure research position in the hard sciences to a programming position in support of research. At the time, I received similar advice from a senior researcher. It was a little more strident though, something like: "What are you nuts? You won't ever be able to propose research again and no one will ever take you seriously."
This is sometimes true, but it depends on what exactly you do and who you work with. I moved from doing molecular biology research (as a PhD student) to writing software in support of same. I have far more exposure now than I was ever likely to get by doing my own research, and I have lots of other researchers (both junior and senior) constantly asking me for help. As a result I've been able to rack up enough publications and visibility that I don't think I'd have a problem moving back to pure research. However, as long as I'm doing methods development, it would be very difficult to get a tenure-track faculty position; I'd basically have to demote myself back to postdoc and do more basic research for a while. Fortunately I have no such delusions.
The bigger problem for the submitter, as others have implied, is the lack of a PhD; this is always going to limit his (?) career advancement.
Comment removed (Score:3, Insightful)
Re:yes, there are a reasonable number of positions (Score:5, Insightful)
Sure, if you define 'humanity' as 'the leeches at the top of the org chart who don't do any actual work'.
An oversimplification. You're not wrong, but the problem isn't that cut and dried. It depends on how they turn a profit. If they sacrifice long-term viability to make the quarterly statement look better, yes, they're turning a profit, but eventually the bad choices will catch up with the company - but by then the people who made the bad choices have long since pulled the ripcord on their golden parachutes and left the rank and file out of a job. I would argue that a better business model is not only to turn a profit, but give the employees a stake in the company's success beyond "you get to keep your job."
Re: yes, there are a reasonable number of position (Score:0, Insightful)
And they go well with the employees who are also money grubbing bastards who will screw the company at any opportunity, no matter the industry.
The lesson here? Everyone is out for their own self interest, its purely natural and there is absolutely nothing wrong with it. Unless youre some kind of religious fucknut.
Re:most definitely yes (Score:2, Insightful)
Hi,
I work in pharmaceuticals. My experience is of the working in the UK private sector in computational biology and synthetic biology.
I don't think ODE modelling is on the way out. It's the standard mathematical tool for modelling in science and engineering. I meant that within my field, roles involving building ODE models of biological systems are not readily appearing on jobs boards. Pharma is interested in probabilistic modelling for clinical trial design as well as pharmacokinetics / pharmacodynamics ("DMPK"), using data from animal and human models. This data is typically low numbers of replicates and very noisy, so filtering the noise and predicting the drug effect, or the optimum patient group to target with the treatment is a trick requiring techniques such as principle component analysis and mixed-effects modelling, as well as stochastic modelling (stochastic modelling skills are also useful to migrate into the financial sector). If you look up "mixed model" on wikipedia, there's a page on this, as well as principle component analysis. Nonmem is the industry standard for mixed effects modelling, or R for academia. Building software around nonmem would be quite lucractive in my opinion.
Small scale models are built with ODE's and PDE's and the parameters are fitted to experimental data sets (non-linear optimization, coding algorithms like particle swarm and genetic algorithms is a handy skill to have. Matlab is a bit rubbish out of the box, although there's plenty on the code repositories). Currently, omics data is very fashionable, measuring all the gene and protein expression levels simultaneously. Whole genome models are popular in systems biology, using linear algebra to build matrices of thousands of interactions, and using techniques such as flux balance analysis (also on wikipedia) to understand how reaction fluxes flow through these huge networks. But this is largely in academia. There are more jobs using flux balance analysis and these "big models" than those building small, concise models of particular pathways, and these tend to part of more wet lab experimental projects. Petrochemical companies are interested in metabolic engineering for biofuels and fermentation processes.
In systems biology there are jobs in modelling, and also software development for the modelling tools. Tools such as the Cobra toolbox for Matlab, Copasi, Cytoscape, SurreyFBA, and the systems biology workbench are examples of these tools coming out systems biology. Trey Ideker's lab at UCS has commercialised modelling tools forming start-ups and consultancy firms. Consultancy companies such as the RES group in Boston have software developers as well as modellers working together to provide modelling and optimization services to all kinds of industries. InSysBio.ru in Moscow are also very good. There aren't many of these in the UK that I've seen (although there might be some in Cambridge), and many modellers I've met are heading for the clinical research organizations (CRO's) and the clinical trial design work (e.g bast inc ltd in Leicester).
In the UK, clinical trial design is most lucrative in the drug company area. Big pharma seems to be breaking up and contracting out particular computational jobs like stats and modelling, and small start-ups are hoovering them up. This kind of modelling leans heavily on statistical modelling though, so might dependant on how much of this you did in your computer science training.
Has this expanded any more?