What areas of IT are hot right now?

[Richdog]

seems a bit odd to specify arbitrary years of experience for a qualification - if their assessment methods are good then they shouldn't need to worry

it isn't like everyone's experience will be at the same level, intensity or that those individuals are as capable as each other

Eh? It's not odd at all, and there is little arbitrary about it. It is a standard feature of most professional certifications and obviously stops people with no relevant experience amassing certifications and watering down their value. If you know someone with a CCISP has 5 years of relevant experience to match the complexity of the subject material contained within the certificate then clearly it's a positive thing.

 

[randomshenans]

You have to have some measure. I get that someone with 3yrs exp could be better than someone with 7+ but you kind of need to have some sort of line.

 

[dowie]

Did you receive my trust message I sent you about 10 days ago?

ah no I didn't but have now, thanks

(sorry had changed the e-mail in my profile but didn't forgot about updating my 'trust' e-mail too)

It's down to the legalities and regulatory aspects - plus add a low risk appetite that means the traditional body of existing banks want to know everything about it before they engage in it.. but ML - that's the point you don't have the answer and qualified risks (costs, impact to the bank due to publicity etc) before you engage in it.. to a certain level you can do pattern recognition, but when you say you're going to use it to discover without that.. that when things get interesting. Especially when you're interacting with customers at the same time

Like Tingle - I'm in a very large financial institution, doing realtime and big data analytics.

I would have thought that argument would apply mostly to areas that aren't fully understood like deep learning. Regulators aside I wonder if the size of the organisations and simply getting approval from senior management is a factor here too?

Certainly some of the people from HFT firms/marketmakers I've spoken to seem to be using aspects of machine learning in a variety of areas such as latency analysis, machine readable news etc..

It's the Nasa scientist vs engineer argument.

One is interested in pushing the boundaries and to take theory & data to propel and test new ideas.

The engineer is there to cope with taking that and making it happen- both at scale and coping with the environment.

I think I understand what you mean - effectively making the scientist-engineer like DevOps, however the scientist needs space/time to spend doing the non-engineering part of building into the existing services. Especially in such a large organisation. The result is that tools like Hadoop, Spark etc are too slow..

Interesting, I mean the traditional 'quants' in banks don't seem to have the luxury of avoiding the engineering side of things, though 'quant developers' exist a quant would generally be spending a large portion of time coding in C++ too. But you seem to be indicating that a 'data scientist' in a bank can do that - presumably stick to prototyping in matlab, R etc.. and not have to get his or her hand's too dirty with the engineering side. That could certainly make for rather a nice role to have Though in general there do seem to be plenty of data science roles out there that will involve some exposure to the engineering side too and do seem to ask for experience using certain tools/frameworks etc..

 

[dowie]

Eh? It's not odd at all, and there is little arbitrary about it.

It is very arbitrary.

Frank has an undergrad degree in mathematics from [Oxbridge] and MSc in Information Security from [some top university] where he did his project/thesis with [big name tech firm] working on some cutting edge encryption related project. Throughout his undergrad he contributed to open source projects and is a bit of a linux geek.

Frank has now been working at an innovative security related startup for 3 years, puts in 12 hour days and is deeply involved in various aspects of the business.

Bruce didn't go to university, spent some time in 1st and 2nd line support then moved to a security analyst job at some big consultancy - his role was somewhat siloed, sometimes he just ran scripts on client sites and sent the results elsewhere. After 4 years he then moved to a security related OPS team where he stayed for another 4 years performing some rather mundane tasks and keeping an eye on some monitoring tools.

Bruce has 8 years experience doing security related grunt work in two different fields, Bruce can get a certificate. Frank is super switched on with a deeper understanding of various fields that aren't accessible to Bruce as he lacks the mathematics background but Frank doesn't have the arbitrary 5 years experience.

The certificate providers realise it is just a multiple guess exam that can be studied for and passed within a few days so they set an arbitrary condition re: work experience. If they had better assessment methods then they'd perhaps not need to do that.

 

[Richdog]

It is very arbitrary.

Frank has an undergrad degree in mathematics from [Oxbridge] and MSc in Information Security from [some top university] where he did his project/thesis with [big name tech firm] working on some cutting edge encryption related project. Throughout his undergrad he contributed to open source projects and is a bit of a linux geek.

Frank has now been working at an innovative security related startup for 3 years, puts in 12 hour days and is deeply involved in various aspects of the business.

Bruce didn't go to university, spent some time in 1st and 2nd line support then moved to a security analyst job at some big consultancy - his role was somewhat siloed, sometimes he just ran scripts on client sites and sent the results elsewhere. After 4 years he then moved to a security related OPS team where he stayed for another 4 years performing some rather mundane tasks and keeping an eye on some monitoring tools.

Bruce has 8 years experience doing security related grunt work in two different fields, Bruce can get a certificate. Frank is super switched on with a deeper understanding of various fields that aren't accessible to Bruce as he lacks the mathematics background but Frank doesn't have the arbitrary 5 years experience.

The certificate providers realise it is just a multiple guess exam that can be studied for and passed within a few days so they set an arbitrary condition re: work experience. If they had better assessment methods then they'd perhaps not need to do that.

I'm happy to disagree rather than waste time arguing something so fruitless. It is what it is, and it is obviously a system designed to fit with the mainstream, and not take into account all individual circumstances that may present themselves.

 

[NickK]

Regulators aside I wonder if the size of the organisations and simply getting approval from senior management is a factor here too?

Yes - you need to have a remit to challenge, reject and change.

Exec sponsorship is a given.. but even with a mandate it's never easy as people all attempt to sabotage and maintain their power within the organisation.

You have two options:
a) create a new legal entity owned but with completely new staff.. bypassing the problem, solve the problem then move the customers over..
b) top down mandate for a unit to create change within the existing entity.

No easy option - both have their pros and cons.

 

[NickK]

But you seem to be indicating that a 'data scientist' in a bank can do that - presumably stick to prototyping in matlab, R etc.. and not have to get his or her hand's too dirty with the engineering side. That could certainly make for rather a nice role to have Though in general there do seem to be plenty of data science roles out there that will involve some exposure to the engineering side too and do seem to ask for experience using certain tools/frameworks etc..

The problem is that an engineer will often look to solve the issue.. Spark/Hadoop etc is just another tech (the argument).. but they miss the point - this is not tech.. it's data. A very different beast.

 

[dowie]

I wasn't talking about 'Engineers' per say but rather data scientists having to do some 'engineering' too. Or apparently not having to do that in some instances and just stick to the 'science' part.

 

[NickK]

I wasn't talking about 'Engineers' per say but rather data scientists having to do some 'engineering' too. Or apparently not having to do that in some instances and just stick to the 'science' part.

So you have data engineers and data scientists.. different from engineers..

 

[dowie]

Sorry but without going around in circles I was talking about data scientists

as I pointed out - on the derivatives side of banking you've got other quantitative professionals in the form of 'quants' and 'quant developers' - though the PhD level physicists/applied mathematicians working as 'quants' don't tend to escape the 'engineering'/ 'quant developer' aspect too.

likewise for data science roles it seems like the statisticians/ML researchers often seem to have to get their hands dirty too

 

[NickK]

Sorry but without going around in circles I was talking about data scientists

as I pointed out - on the derivatives side of banking you've got other quantitative professionals in the form of 'quants' and 'quant developers' - though the PhD level physicists/applied mathematicians working as 'quants' don't tend to escape the 'engineering'/ 'quant developer' aspect too.

likewise for data science roles it seems like the statisticians/ML researchers often seem to have to get their hands dirty too

True- however typically they will start with the initial idea.. often doing some basic engineering as an experiment but taking that to scale is where good data scientists have typically been pulled in the olden days only to regret it.. the modern thing is you have some but not a massive number of data scientists purely dedicated (they exist ) and also other disciplines related to it.

There's a definite move away from attempting to get an engineer doing the scientist role as you've highlighted too.

 

[platypus]

DevOps, yo.

 

[Capt Doufos]

AWS + DevOps. Our company cannot get enough, we are struggling to get the numbers. As are many others in the same space.

 

[Energize]

How does one break into the devops field?

 

[dowie]

presumably persuade your employer to change the way things are run or move to an employer that already embraces 'DevOps' culture

that is assuming you're already a developer or sysadmin person

 

[MikeTimbers]

There are coding acadamies that teach the DevOps mentality. Good schools are not cheap but work with employers to place their graduates. This pretty much guarantees students a job at the end of the course.

DevOps is very hot but most employers don't actually understand the term and use it for a role that does the code deployment rather than for a full-stack developer who also knows how and is empowered to deploy automatically-tested code to Production.

 

[BigBANGtheory]

Don't underestimate the combination of good and relevant technical skills with communication skills. A great strength of the UK workforce is communicating in English to a high standard which to put bluntly is not easily off-shored like Sysadmin roles.

In IT you want to be the person that governs, builds, creates, designs and is in pre-sales not the administrator and career of things.

 

[M3G4UK]

Just to put my two penneth in...

It's not ALL coding. If coding isn't your strong point (it certainly isn't mine), then cloud architecture and cloud experts are quite in demand, especially with a lot of companies moving to virtual stacks either in Azure or AWS.

Networking is always going to be hot as without network infrastructure, a cloud isn't worth anything.