Programming
Tools
- Python, PyTorch
- MATLAB
Curriculum vitae
Kasun Dewage
Ph.D. candidate in Mathematics at the University of Central Florida, working at the intersection of tensor methods, numerical linear algebra, efficient foundation models and quantitative modelling. Expected completion Fall 2026.
Education
Expected Fall 2026
Ph.D. in Mathematics
University of Central Florida · GPA 3.7
Advisor: Prof. Marianna Pensky
2024
M.S. in Computer Vision
University of Central Florida · GPA 3.8
2017
M.S. in Pure Mathematics
Sam Houston State University
2014
M.S. in Financial Mathematics
University of Colombo
2012
B.Sc. in Physical Science
University of Colombo
Appointments and experience
2020 — Present
Teaching Associate / Instructor of Record
University of Central Florida
Calculus II, Calculus III and Linear Algebra; up to 8 credit hours per semester. Student evaluation averages between 4.6 and 5.0.
2018 — Present
Independent Quantitative Researcher
Systematic trading
Systematic trading and risk-management strategies using mathematical modelling, statistical analysis and computational methods; quantitative models for market-regime analysis, forecasting and benchmark-based evaluation.
2017 — 2020
Full-Time Lecturer / Pool Faculty
Sam Houston State University
Up to seven mathematics courses per semester, including College Algebra and undergraduate service mathematics.
2015 — 2017
Graduate Teaching Assistant
Sam Houston State University
2014 — 2015
Banking Adviser
Standard Chartered Bank
Talks & scholarly presentations
Presented · 2026
Hybrid Neural-Classical Correction for Frozen Time Series Foundation Models
IEEE WCCI / IJCNN 2026 · Maastricht, the Netherlands
IEEE World Congress on Computational Intelligence / International Joint Conference on Neural Networks.
Presented · 2026
On the Compressibility of Fine-Tuned Attention: A Tensor Decomposition Study of PEFT Weight Structure
IEEE ICNLP 2026 · Xi’an, China
Expected · 2026
LoRA-CRAFT: Cross-layer Rank Adaptation via Frozen Tucker Decomposition of Pre-trained Attention Weights
COLM 2026 · San Francisco, California, USA
Expected scholarly presentation at the Conference on Language Modeling.
Expected · 2026
Accepted work on transformer compression and structure
IEEE ICMLA 2026 · Rochester, Michigan, USA
Expected presentations covering magnitude-profile pruning, quantization sensitivity, and spectral outliers in transformer attention.
Skills and recognition
Mathematics
Methods
- Tensor decomposition, low-rank modelling
- Numerical analysis, linear algebra
- Statistical computation
Artificial intelligence
Areas
- Transformers, foundation models, generative AI
- PEFT / LoRA, compression, pruning, quantization
- Time-series forecasting
Sponsored research
NSF Award No. 2310881
Significant contributor, Multiplex Generalized Dot Product Graph Networks: Theory and Applications. PI: Prof. Marianna Pensky.
Awards
Outstanding Faculty Award nominee
Sam Houston State University, 2019.
Memberships
Societies
- IEEE
- International Neural Network Society
- American Mathematical Society
The document