February 20th, 2012, 9:45 am
Victor is being commendably honest, not knowing quite what to look for is pretty common in hiring managers.Not least because one is often hiring people to do things that you can't do for yourself.Actually I'm writing a "how not to interview" piece for TheRegister.com and it will contain stuff like:Firstly, forget the idea that you can interview these people all by yourself. If you want him to work with architects then get one to talk to him. The trick here is to ensure you sit in on the chat, you don?t just want someone who can do ?X? a BA role includes a lot of interaction and you?ll want to see that.Rarely will you find the manager of the architects against this, especially if you pitch it as ?he?s going to be working with your guys, so I need to see how well he does that?. Better to get an actual architect than a manager of architects.Ditto, some trader.I assume you have formed some outline model of what you want, but given the time constraints of the interview process it is valuable to work out what it is you don?t want, dealbreakers make interviewing easier and reduce the chance of you getting blamed for hiring someone grossly unsuitable.That?s for you to work out, but one example comes from my experience of BAs in banks is that often they have far too little understanding of technology, I define that as seeing traders end conversations with BAs very abruptly when they realise that the BA has vastly less tech knowledge than they do, to quite one managing director at a very large bank to the division head of IT, ?why do you hire these clueless fucks ?, get some programmers in?.A good BA is not a passive recipient of demands from ?the business?, he interacts with things like saying ?we could do this risk calc in real time but that would cost serious money, but having it spit out the current exposure every 10 minutes is almost free and can be ready much more quickly?. The 2nd worst sort of BA is one who cannot be in the loop because he always has to ask a tech person to produce any useful estimate on times and costs and certainly can?t make suggestions. The worst of course is one that promises and cannot deliver.I?d only ever consider hiring a BA who had been a programmer for some useful % of his career and as a headhunter that would still be my filter. That makes tech questions relatively easy to find, ask a former C++ programmers about smart pointers, a Java one about handling garbage collection in a real time system, an Excel/VBA about calling DLLs, SQL about outer joins, Btrieve about fixing indexes, FPGA about reducing the bulkiness of automatically generated systems, etc.Of course he also needs to have quantish knowledge, a good quick question is asking about mean reversion in FX. This is a question where you can answer it shallowly (ie any rate can go to infinity or zero and thus cannot be mean reverting), a bit deeper as in the way some rates seem to do that to some extent for a while, to excellent where they describe how you can exploit local MR to make money and how to detect when the MR is changing regimes so you can get out cheaply.A BA isn?t a quant or a developer, he needs to talk to these people with enough credibility that they will work with him, so in this very specific case simply getting a copy of Wilmott for quant stuff, some random programming book for the language they claim to know and asking them ?definition? type questions works to filter BAs usefully. A good sanity test is to ask them what the ?buzz? was in their business unit yesterday, some market related topic. That shows they are in the loop where they currently work, which implies they can do that at yours.