Dr. Kyriakos Christodoulides

Wait, wait, wait! ML interviews need critical thinking? Well, slap my circuits and call me surprised! I always thought those geeks just pressed a 'Make Magic Happen' button! "AI" not so I these days...

Search for honesty, clarity in communication and a combined datascience and subject matter expertise...but only if you can handle being honest and clear with yourself first

a) You can stand out as technically excellent, and in this case in-depth understanding of algorithms, data structures and robust coding practices should be your priority. b) If you want to stand out as a creative problem solver though, useful for blue-sky-type projects, then in my opinion having multidisciplinary experience and a really good grasp of the underlying mathematics is crucial. Being able to communicate and network effectively on top of being either a) or b) is Key. If you also become a domain/industry expert as well, then you are unique. If on top of the above, you also possess the capacity to mentor, teamwork, continuously learn and you are aware of ethical considerations, then my friend you are a unicorn...

You have to always remember that ML problems by themselves are open-ended (else we 'd be able to solve them with an exact science like physics), always admit more improvement and constraining the problem through business understanding is of paramount importance. A solution that is 'good enough' might be fit for purpose. But you do need to be able to quantify 'what that purpose is' The techniques that we use and the 'level of logical reasoning' achieved, should be justified by ROI and tailored to a quantifiable "what success means for the business".

If you are at the beginning of your career, then technical upskilling is probably the way to go about. The further you are in to your career though, upskilling needs to be done at the level of building relationships and a network. This comes from going to conferences, public speaking, writing articles, interacting with communities on LI e.t.c In my opinion leadership courses, strategy courses e.t.c do little to genuinely help unless you leverage them as a way of building your network. The ultimate upskilling is of course to build your own business. It includes all the above mentioned and more...

No matter what software you are using, you will end-up with false expenditure including duplicate invoices. Each case is different, hence exploratory data analysis is required. Without exploratory analysis first i.e an evidence based approach, any hypothesis or even measure you take is meaningless. Then you should use Statistical Modeling and Machine Learning. The solution: Hire a datascientist who specialised on the subject, understands risk as well as the business implications of such a project. Do not auditors already do that? No they don't and they can't.