The Phenoscape project attempts to formalize the description of evolutionary characters to make them interoperable and computable with the body of phenotype annotation being generated by model organism databases, and other biomedical research.
OpenBRR2 project focuses on the re-write of the original "OpenBRR - Data" (http://sourceforge.net/projects/openbrr) project. The aim is to develop a hosted data aggregation application for all F/LOSS projects.
Design and develop Recommendation and Adaptive Prediction Engines to address eCommerce opportunities. Build a portfolio of engines by creating and porting algorithms from multiple disciplines to a usable form. Try to solve NetFlix and other challenges.