Dr Sowmya teaches courses in software engineering and computer vision and learning , pattern recognition and robotics.  These courses are at the third and fourth year level, with software engineering being a core course taken by students B. E. Computer Engineering, while the vision and robotics courses are advanced electives at the fourth year level.

Dr Sowmya's research is in two areas: image analysis and recognition, and software engineering. The first area focuses on feature extraction, recognition and understanding of high resolution images, in particular satellite and aerial images and medical images. Techniques are drawn from machine learning applied to computer vision, pattern recognition and statistical data analysis.  Robotic vision and control as well as visual robotics also form part of this area of research.  Past and current research projects include road network extraction from high resolution aerial images and digital maps, symbolic learning techniques for object recognition, real-time resampling and tracking algorithms  and motion recognition for a security camera and medical image understanding and diagnosis on HRCT images of the lung. 

The second area of interest focuses on formal methods of specification, verification and design of real-time, reactive and distributed systems. Techniques are based on process algebras, temporal logic, simulation and deduction. Past and current projects include  verification of statecharts using logic-based techniques, real-time (robot) control software development using Esterel and design reuse techniques for component-based embedded system development, reuse techniques for Hardware Description Languages such as VHDL.