This site may give you a flavor of the the kind of problems I am interested in, but there is much to add and this is a work in progress. I spent a long time developing software for business tasks in various domains, I find reformulation of these problems in terms of algorithmic or learning tasks enjoyable and interesting. With that said, here is a reasonable summary of my professional interests.
-
Framing business problems as algorithmic problems
-
Framing business problems as machine learning tasks
-
Framing business problems as statistical modelling tasks
-
Doing the above with a satisficing perspective if need be. Optimal is a worthy goal, but efficient learning is a worthier goal. By efficient learning, I mean learning with time and costs that are just enough to meet the business need.
-
Framing big data problems in terms of newer algorithmic methods such as submodularity and sketching (see this playlist ) so that large problems can be solved with sufficient (satisficing at work again) solution quality with a fraction of the costs and resources. I really believe in adding complexity only if needed, but definitely if needed.