Carol V. Alexandru

Research Assistant
University of Zurich
Department of Informatics
Binzmühlestrasse 14
CH-8050 Zurich, Switzerland
Email alexandru@ifi.uzh.ch
Office BIN 2.D.14

About myself

I obtained my Master of Science in Software Engineering at the University of Zurich on the fast track in Fall 2014 and have been pursuing my PhD since.

My research currently centers around the question how we can leverage the vast amount of heterogeneous source code available on platforms such as GitHub and Bitbucket using automated, large scale approaches (as oposed to traditional, query-based methods). Of particular interest is the question: Can we teach computers how to program? Using existing source code as training data for machine learning, researchers are able to create models that can aid in writing new code, fixing bugs, translating and de-obfuscating source code.

Beyond this, "big data" code repositories have yet to be fully exploited on a large scale in areas such as empirical studies on software evolution, improving automated software quality assessment approaches and search-based software engineering, just to name a few.

My goal is to develop methods and tools which allow us to tap into the enormous, combined knowledge contained in existing programs and re-use this knowledge for new purposes.

Previous Projects

Facets of Software Evolution

As part of my bachelor thesis, I developed an easy-to-use web front-end for running and viewing in-depth software quality analyses called Facets of Software Evolution.

seal Facets

 

Signal/Collect Console

During my master project, I developed a debugging and introspection tool for Signal/Collect, which allows users to visualize the graph they are working on and step through the graph computation one step at a time while observing changes in the graph. The Introspection tool has been integrated into Signal/Collect.

seal sc

 

LISA

During the fast track MSc I developed LISA, a high-performance, language-independent library for formulating and running static software analyses on projects stored in Git repositories. The main advantage of LISA over traditional code analysis tools is the ability to analyze thousands of revisions in parallel, as opposed to one release at a time.

seal lisa

 

Teaching

Informatics I (Assessment) (2013: teaching assistant, 2014, 2015, 2016: course administration and lectures)

Publications