Tomasz Melcer


Tomasz Melcer is an expert in solving problems that stretch across software engineering, machine learning and system administration. He has 10 years of commercial experience, including implementation of classification algorithms for cancer biomarkers detection, development of software for mass spectrometry data analysis, and highly scalable systems for monitoring and predictive analysis of computer networks. A supporter of open source software, he participated three times in the Google Summer of Code program as a mentor for projects involving natural language processing and R libraries. Tomasz is currently working at Wroclaw University of Science and Technology, where he conducts research on machine learning methods in diagnostics of eye diseases. He also works with QuantUp, where he provides trainings and supports commercial analytical projects.



Software Engineer in the Machine Learning world (2018)

For a software engineer, the code is a medium for building applications: modern tools that help millions of people in their daily lives and work. The prime goals are stability, compatibility, human interface, etc… But for a machine learning practitioners, coding is a tool for preparing numerical experiments. Their code is often executed only once and its requirements always change. Buggy code's output may look perfectly fine, yet be severely misleading. And if the code gives a "wrong" result, it does not have to mean the code is wrong! The challenges are different, and their solutions often unique. I will show my story of entering the machine learning field as a software engineer. I'll peek into the culture of machine learning programming and focus on what is different, what is not obvious or even surprising, what are some ways of solving problems peculiar to this quickly developing field.