Helping taking the research to a more robust process and moving it towards production
the purposes were:
Establishing the research results and validating them
Converting the code into a more manageable and modular code
designing and executing of a process built on the work done and teaching the in house data scientist how to do it.
my tasks:
To challenge and validate the former research results, converting the code to a more robust and performance aware architecture and moving it towards a service, while teaching the in house data scientist how to do it.
challenges:
Mainly lack of data, need to infer by features we did not have in the training set, and lack of clarity about the missions and priority
I chose the technologies.
difficulty:
lack of clarity regarding the priority and contradicting missions given by different persona in the company.
Stack:
Bitbucket, ubuntu, python (sklearn, numpy, pandas, flask), tensorflow, keras