Upon completion of the Bachelors degree in Mathematics, Thashmee shifted her interests towards Statistics and Computer Science, and later purely into Computer Science. She persuaded her work towards Artificial Intelligence (specifically expert systems) with a Master of Philosophy degree, which was based on three years of full-time research. At the same time, she was occupied as a university teacher, taking a number of computer science courses at the undergraduate level at the Open University of Sri Lanka. Her interest in the field of machine learning has emerged when she started her Ph.D. studies at the Department of Computer and Systems Sciences (DSV) of Stockholm University, under the supervision of Professor Henrik Boström. She continued teaching at the Open University of Sri Lanka after completion of the Licentiate degree at DSV. Thashmee conducted research towards the Ph.D. from 2009, until 2013, and obtained her Ph.D. in March 2014.
During her research studies, she has worked primarily on how to learn efficiently and effectively from COMPLEX and STRUCTURED data. Her approaches included representing complex data as mathematical graphs, and use graph mining methods to discover important subgraphs where machine learning methods are used to learn from them. She focused on transforming graphs into simpler structures, i.e., Itemsets, which allow the use of itemset mining methods to learn from graphs. The simplicity of this approach has resulted from increased efficiency by the less computational complexity of the methods thereby allowing the methods to be applied to large graph databases. She has applied her methods successfully on different domains including Medicinal Chemistry and Web Databases. Most of these research are available in several peer-reviewed publications.
She is currently continuing her research in Educational Data mining and learning analytics and in the ICT4D group. The learning analytics research includes visualization of educational data in different aspects in order to overview about many indicators of learning with respect to a target student group. Discovery of information from user logs (digital traces) of online systems, which could be useful for finding ways to how to maximize the learning and how to design suitable pedagogy for better learning outcomes is the main focus. Investigation about what knowledge can be discovered by analyzing data in online repositories may be useful to a) students for their motivations in learning, b) teachers for improving pedagogy and course delivery, c) administrators for improving efficiency and the effectiveness of their services, and d) the department for decision making and policy improvements. In these studies, the focus is to investigate how and which ways the quality of education can be improved by the knowledge extracted from data accumulated in the repositories as a result of using automated systems.
In line with the collaborations with many countries in Africa and Asia as a project manager, she is also investigating the impacts of the use of ICT in many contexts. These works include how ICT can support collaboration and communication among different stakeholders in different setups, such as education and entrepreneurship for example, for increasing the throughput. Also, she works on how to tackle high dimensional data, especially reducing the dimensionality using itemset mining approaches. Learning patterns from large graph datasets which are categorized into disproportionate clusters is another question she is currently working on.