Functional programming (FP) has become the New Big Thing in finance over the last 5 years or so. This shift means that I get a lot of spam from recruiters, as somebody working on a compiler for a FP language. Most of these e-mails are tailored or at least have my name in them and a basic acknowledgement that I’m probably going to finish my Ph.D. before I do anything else. Which seems like a reasonable amount of effort from the recruiters, given that these are 250k+ jobs and they take a sizeable chunk in commission.

But GS? Well, nothing says, “we’re interested in you as a person” like a To: line of undisclosed-recipients. And a body completely devoid of any mention of interesting work, just that the Stats desk is hiring and they are looking for people who work on compilers for functional programming languages.

Especially since then I realized: that’s maybe 5-10 people in the US (there are ~10 semi-active FP compiler projects, but most are just a single faculty member and no NSF funding or grad students. Europe and the Pacific region are much higher, naturally.).

Seriously? You could customize mail for that many people in about four hours, tops. The Stats recruiting team is probably larger than the entire set of in-training talent. As somebody who has spent a lot of time on the hiring manager/recruiting side of things, I was sort of shocked. Given their size and position in the market, I was expecting them to compete with the aggressive, personalized recruiting I’ve been seeing from Credit Suisse, JP Morgan, and just about every HFT shop. Not that I need any more recruiter spam!

Unless, of course, it’s for research universities that are looking to hire someone to teach and do research on compilers. I’ll take as much of that as is going around — the academic (particularly state school) job market is  brutal here in the US.

In practice, they screw up all the time. From the obvious, where they demand more years of experience in a technology than the tech has existed to the less-obvious mutually exclusive skill sets, I’ve seen and given talks about many of the mistakes hiring managers make. Rather than make a separate post about good vs. bad job postings, though, I will instead poke holes in some random non-confidential job postings that come across my inbox, particularly via mailing lists. All company names will be removed, naturally.

The following just hit my inbox. What’s wrong with it?

Job Description
Elided, a quantitative trading firm based in Chicago, has a job opening for an expert developer. The person would be responsible for developing and enhancing trading systems including exchange connectivity, overnight jobs, and databases. Supporting and maintaining production systems for several geographical markets, and participating in systems and storage configuration and maintenance.

Requirements
BA or MA Computer Science, Mathematics, Statistics, Engineering or related field; Candidate must be an expert in C++ and Linux, and competent in: Visual Basic, scripting (bash,awk,sed). Knowledge of tcp/ip, VLANs, 10G ethernet, Postgresql, SQL Server, RAID, clustered storage, and remote hardware deployment will be helpful. No financial experience necessary.


This was sent to a jobs mailing list for UChicago CS graduates. So, that’s either a 4-year undergrad, a Ph.D. (or drop-out master’s), or a targeted master’s student — which is usually someone without an existing strong CS background. How many things can you find wrong with this post? Here’s my list:

Major show-stoppers resulting in zero candidates:

• Visual Basic. Required. It’s rare that I see a job advertisement reduced itself to < 0.1% of the graduating students from every top university in a single shot.
• SQL Server, for a new college graduate, in this day and age. I’m not even going to comment further. And my wife worked on the SQL Server team for three years.

Minor annoyances:

• Expert developer implies many years of professional experience, probably leading a multi-year project with junior developers working alongside you on smaller portions of the project. How many of those do you think are about to graduate? And further, do you think they even need to look for a spammed job?

That said, I’ve certainly seen worse! And, in future blog posts, I’m sure I’ll cover them, too…

Before returning to graduate school to work on my Ph.D. in pursuit of creating a gigantic, compilers-focused research group, I worked at Microsoft. I was a developer and manager of developers (and testers) in the Developer Division, which makes Visual Studio and a bunch of other tools for software developers. In that time, I performed well over 200 interviews of new college graduates. Some were part of a day-long loop, but many were also half-hour screening interviews during 8-10 hour days on college campus during those “interview days” the larger career centers put on.

While your mileage may vary, most of the advice I will give is based on real situations and mistakes new college graduates routinely make. In the talk version of these notes I provide colorful anecdotes of real-life Darwin Career Award candidates. Unfortunately, in the infinite-life web version, I’m reluctant to provide them, even without names, as I give them in the talk. While all candidates I rejected from interviews at MSFT were rejected based solely upon their ability to perform the job at hand, I’d prefer to avoid any legal conversations.

In this post, I discuss resumes. Over the next couple of posts, I will also cover interviewing tips and actually landing an interview in the first place.

Resumes have two purposes: the first is to get the interview, and the second is to give you something to talk about with your interviewers.

Getting past obvious screening

Except in the cases I’ll talk about during the discussion about landing an interview in a later post, you should assume that your resume is being read by a cruel, hung-over reddit grammar fiend. Most college students imagine that their resume will go from their editor into a generic e-mail address or web form and then onto the desk of a technical hiring manager who will then make a decision about them. At all but the smallest firms, that couldn’t be further from the truth! You know that jobs@company.com alias? It gets hundreds to thousands of e-mails. Each day. I can only imagine things are worse now, given the job market and incredible increase in CS enrollments over the last few years (which I couldn’t be happier about!).

Skills

At MSFT, the recruiting department had specific tools to recognize “chain submitters” and filter them out. But even lacking that, most companies will end up putting either human resources or a dedicated recruiter on the front lines, filtering resumes to a more manageable level. So, what do you have to do to get past them? To be honest, this is where that ridiculous skills section comes in. Take a pass over your resume and ask the following question:

If someone at the company you wanted to work was scanning your resume, but had no idea about how the products were built other than the basic names of technologies involved, would they bring you in?

So, even assuming you have a CS degree from a good school and a 3.8/4.0 GPA, if they have 25 other candidates from similarly good schools with similarly good GPAs, do you have even a passing familiarity with their tools? If they’re a Unix shop and you only list Windows-based tools in your skills, that will probably prevent you from getting even a phone screen.

It’s not that they don’t think you’re good. It’s that they only need to hire a couple of people, and due to the number of applicants to competitive companies, they can afford to be selective.

As an example, until recently, I thought that all of the Stanford CS kids did the five year dual Bachelor/Master’s degree program, but I have since learned that many of the BS-only students couldn’t even get past the on-campus screen.

No typos or grammar errors

Seriously. It’s one page. Get someone else to proof it closely in exchange for checking theirs and be very careful every time you change it.

Don’t give anyone along the way a stupid, petty excuse to reject you.

Have something interesting and CS-related

Particularly for engineering schooles (due to their accreditation-friendly curriculum), everyone took the same classes. With the same content. And basically did the same projects, since there are just a few books catering to each of the mid-to-upper-level courses.

How are you different?

Find that, and point it out. Hopefully it’s CS-related and involves a significant hacking project of some kind (see the later section on driving the conversation), but even if it’s a background in statistics that has made you particularly good at rigorous performance profiling, call it out. You need to find a way to make your resume stand out from the rest of the ones in the pile, and cinnamon-cookie scented paper isn’t the trick.

Well, that would’ve worked for me, but you shouldn’t count on it in general.

Miscellaneous content

One page is an ideal length for a new college graduate’s resume. If you have so much CS content that there’s no room to put on your crew team, debate club, and residential college leadership roles, omit them. They won’t be missed. As a rule of thumb, that also applies for your first few years in industry. Your resume will see its college section shrink down to just the major research projects you participated in (if any) and that space filled with your first job or two.

Don’t add unnecessary things. Gap in your schooling due to personal issues? Leave it out unless it adds something to the conversation (i.e. you traveled the world while using your custom geo-tweet app every 500 paces taken). No hiring manager I’ve ever met will change their opinion of you based on whether it took you four or six years to get through school. And if you’re worried, and anyone asks during the interview, just reply, “personal issues, which are resolved now.” That, by the way, is interviewee-speak for “if you keep asking me questions along these lines and later refuse to provide me a job, you’ve given me grounds to retire off of the legal settlement.”

Drive the conversation

The number one reason recent college graduates are subjected to stupid questions is that they failed to provide something interesting to talk about. And it’s a true shame. Your resume is a tool to drive an interview. If you list no projects? Stupid questions the interviewer makes up. If you list some open source projects or research you participated in, they will ask you. In an uncoordinated interview, you may even be asked several times.

So, first, get on a project! Got Federal work-study money? Go talk to a professor from a class you did well in and tell them you have it and want to work on something for them. You’re free labor, and any (particularly systems) professor worth their salt has a nearly infinite supply of undergraduate-sized projects available. And if you don’t have work-study? Offer to do the work for credit. You will learn more fixing the network interfaces for the basis library of your local Standard ML implementation than you will in even a graduate-level networks class.

Further, you’ll have something to talk about. So, be prepared! Imagine that it’s the first time seeing your resume, and the interviewer will say, “tell me about your work with the networking libraries in SML/NJ.” You should be prepared to describe the overall project, how your code fit in, what you fixed, what you added, and what was remaining. Information about how the code is used by callers is a bonus. Trust me, being able to do that will separate you from the other candidates, most of whom give one of the same two sob stories: “I worked on a group project in my databases class but I was the only one who did any work” or “my courses did not have any large hacking projects.”

If you can, make sure you have at least one deep technical bug or persnickety detail in your mind. Odds are decent if you have something such as a networking-related project that a great company will schedule your interview loop so that one of the random members later in the day will be the primary author of WINSOCK or the person who put together the FTP support for FireFox. Talking about that detail, how it threw your for a loop or broke your code, and how you finally tracked it down should lead to a great conversation (or friendly argument!). A vote of technical confidence from people with that kind of background can easily wipe out a mistake you made earlier in the day when implementing newspaper line justification on a whiteboard.

BONUS: Own your online presence

Assume that interviewers will use Google and Bing to search for your name. Own the first page of results for you. Both of those engines allow you to request removal of pages you own (honored within two days).

Most dubious content cannot legally be used to make a hiring decision and it won’t be by large, well-trained companies. But, do you really want them asking who’s got the “weaboo DDR wizard” next?

But probably the largest positive lesson I’ve learned from this process? Get involved in grant writing — especially as a graduate student!

Best early feedback you will ever get

Sad but true: when you are proposing your Ph.D. thesis topic, it is very hard to get even an involved committee to think hard about issues related to your problem area. If only there were a peer-reviewed, important publication venue that you could use to as a sounding board for your idea, plan, and coverage of the related work…

I received more review of my early stage ideas during the grant-writing process than I probably ever have or will again during my academic career. And the grant reviewers are often drawn from a broader set than your normal conferences (organizations such as the NSF desire “broader impact”), so the feedback you will get from some of them will also be unique.

If you are planning to become an academic, it’s a critical skill

Research money supports not only your research, but also the university (through the > 50% of overhead charged against grant funding) and eventually your tenure case. Grant proposals are unlike any other form of writing you have probably done in the past. The NSF’s “Broader Impact” requirement includes benefits to society. For an area similar to mine — programming languages and compilers — thinking about and learning what that means takes time.

Also, unless you subscribe to the “write a grant for the thing you just did to fund the work you will do” camp of grant funding (my advisor does not), this proposal will be one of the only documents you write without concrete results backing them up. It is a balancing act to put out just the right number of details. Too few, and your planned research will not convince a funding agency that you know what you are doing. Too many, and you open yourself up to criticism for making choices that clearly will not work according to the reviwers.

It helps the group

By the time you can help with the grant-writing process, you will probably be in your fourth year of graduate school or later. If you are like me, you took advantage of the existing grants, resources, and people. Helping to obtain additional grants — even if you will not be around to use part of it — can only make the project better. And what better way to make your work appear more important than to have the project continue to flourish after you leave?

Machine configuration
Connect a standard ethernet cable between the two machines.

On each machine, bring up the list of network adapters by opening the Start
menu, searching for View Network Connections, and then clicking on the Control
Panel item that is shown. Right-click on the Local Area Connection connected to
the cable and select properties. If you have multiple Local Area Connections, it
will be the one labeled “Unidentified Network.”

Double-click on the Internet Protocol Version 4 (TCP/IPv4) item. In the dialog
that comes up, select the Use the following IP address radio button. There will
be different settings for the two machines.

Machine A
IP Address: 192.168.2.1
Subnet mask: 255.255.255.0
Default gateway: 192.168.2.2

Machine B
IP Address: 192.168.2.2
Subnet mask: 255.255.255.0
Default gateway: 192.168.2.1

Click the OK button until properties are set. At this point, the two machines
should be connected!

Using Easy Transfer

Just launch Easy Transfer as normal, using a network connection. The two
machines will be able to see one another. To ensure you do not accidentally go
over your home’s local, slow WiFi network, you may want to disable the Wireless
Adapter on one machine. To do so, from the Start menu search for View Network
Connections. Right-click your Wireless Network Connection and select Disable.

Copying files

On the target machine, create a network share by right-clicking the folder to
share and selecting properties. Then change to the Sharing tab and click on
Advanced Sharing. Check the ‘Share this folder” box, then click on
Permissions. Click on Everyone and then Remove. Click Add, type your user
name in the box, and click OK. Select your user name, click Allow/Full Control,
and then click OK until all of the dialogs are dismissed.

Connect to the share by entering the following command:
net use * /persistent:no /username:TARGET\USER \\192.168.2.1\SHARE

Remember to replace TARGET with the name of the machine, USER with the username that you gave access to that share, and SHARE with the name of the share form the dialog. A drive letter will be assigned to the network share, and you can now quickly
browse and copy files between the two machines (Z: unless it is already in
use).

Professional debt

Each paper is reviewed by 3-5 people, depending on the conference. So, if you submit a paper, you accrue a debt of 3-5 papers (admittedly, split with your co-authors). As a young graduate student, you can’t provide concrete feedback. Even at my stage, later in my graduate career but still a year or so away from completion, I can only provide reviews in a few very narrow areas. Given the number of people who will submit, graduate, and never pay off their debt, just keeping the whole research publication game alive requires you to review quite a few per year!

Programming languages reviews are fantastic

It always impresses the other University of Chicago graduate students in areas like supercomputing and theory when I tell them about the quality of reviews in PL. Overwhelmingly, every review we’ve gotten for Manticore has been thorough, deep, and insightful. Not just one long and detailed response per submission, but nearly every individual review, from major conferences to workshops to journals to even NSF grant proposals! I enjoy attempting to meet that high bar set by my peers and find the kinds of persnickety technical details, key missing related work, and insights about wider applicability or opportunities for extension that we’ve been given on our project. Seeing my review next in its place next to 2-4 other reviews of a paper can be humbling, especially when I’ve missed a glaring error or come away with a different position on the paper than all the rest of the reviewers.

Really understand the submission

Short of implementing the meat of a paper or reproducing its proofs by hand, writing a thorough and constructive review is the best way for me to believe I really understand what the paper is presenting. What does it build on? Do I believe its claims? How did they evaluate their results and is that a meaningful measure of their claims?

I’ll take the express mail international for $46. Can you send me a paypal money request for the shipping price plus the buy it now price? I did a lot of shopping last friday and I just wanna make sure that I have enough to pay you. My paypal is UnrelatedToEBayUserName@yahoo.com thanks!

For those not in the know, eBay Seller (and Buyer) Protection only happens if you perform the purchase through the site. As soon as you go directly to PayPal, you’re out of luck. They can use either a hacked account, whose purchases will revert when the hack is finally investigated, or just dispute and cancel the payment. You have no recourse. After requesting the purchase directly through eBay, he went away.

Later, I got mail that the item had sold! But wait:

Hello Seller How are you. I am XXXXX i am contacting you because of your item that i won on ebay and i want it shipped to my son who is in west Africa. He is doing his birthday next week. i will be responsible for the shipping cost and i will be paying you through paypal. if you are ok with it get back to me with the invoice and your PAYPAL ID Here is my SemiRelatedToEBayUserName@yahoo.com so that i can make the payment to your account once, I will provide paypal the shipping address once i pay for the item.

I would like to have the item shipped to the address below:
XXXXX XXXXX
MOBOLAJI BANK ANTHONY WAY
IKEJA, LAGOS 23401 Nigeria

Seriously? Nigeria?

EBay has certainly done one positive thing — my Craigslist experiences now seem like a fraud-free dream.

We continue to scale well past 16 cores

We have shown in the past that Manticore scales very well on a variety of benchmarks up to 16 cores. But, this time we’ve been able to show that we continue to see speedups all the way to 32 cores! Of course, there’s also bad news. While we have decent speedups at 16 cores and admirable speedups (relative to competitive implementations of the chosen benchmarks) at 24 cores, our 32 core speedup is only slightly better than that of 24 cores on some benchmarks. We need to do better. And more, we need to figure out where the bottleneck is!

We had a huge GC bug lurking, more than a year after our last one

Our benchmarks and output in general have gone through some very heavy stress testing, on decently sized manycore machines. However, whenever you double the number of cores, you can expect to find something new. And in this case, on one of the 3 machines, SMT is enabled, allowing 64 cores of testing. And at load… there’s a garbage collection bug. Three days of groveling over the heap, generated assembly, and instrumented binaries enabled me to work around it, but one of the downsides to a highly parallel garbage collector and pretty complicated scheduler is that when there’s a bug, it can take a while to track down the real root cause.

SMT does not appear to be useful for high-computation loads

At least for the sorts of work that most of our benchmarks have, SMT (which doubles the number of available cores by allowing two threads of work per physical core) does not offer significant advantages. Of course, this will need more investigation. Right now, when we spawn up additional threads, they each need their own nursery for garbage collection. But, we don’t resize those nurseries so they will all fit within the L3 cache size, potentially causing some memory thrashing that isn’t evident without the additional SMT threads. I also suspect the poorer speedups at lower numbers of processors is due to densely packing processes rather than packing densely at the package and core level but using extra SMT cores last.

Want to learn more? Well, this and other results are the topic of a large journal paper we’re writing on our work building the Manticore runtime system, which was specifically designed to take advantage of manycore architecture machines. Of course, in the journal paper there will be real numbers, rudimentary statistical analysis, and concrete recommendations. This article is just a post to thank the fantastic folks at Intel for making the Manycore Testing Lab available. I highly recommend it to anyone in an academic setting who doesn’t have an extra $30k to spend on one of these machines. Well, after I finish slamming those machines with work myself, of course!

And, it certainly won’t hurt to discuss various persnickety details of Manticore, our parallel mostly-functional programming language and runtime, that don’t merit publication.