My research spans medical image analysis, artificial intelligence, optimization, and large-scale system design. I work across academic and industrial contexts, focusing on building models and systems that are not only performant, but also interpretable, deployable, and aligned with real-world constraints.
My doctoral research focuses on automated interpretation of non-contrast brain CT for acute stroke and intracranial hemorrhage assessment. Emergency clinical workflows demand fast, consistent, and explainable decision support under significant time pressure.
I develop deep learning pipelines—particularly Vision Transformer–based architectures—for stroke classification, hemorrhage detection, and related clinical tasks. Emphasis is placed on comparative evaluation, dataset curation, and understanding model behavior rather than optimizing isolated benchmarks.
This work is informed by applied R&D experience in production environments, ensuring that methodological choices consider scalability, robustness, and system-level integration.
Vision Transformer–based stroke classification framework enhanced with GAN-driven data augmentation to improve robustness on limited clinical datasets.
Comparative evaluation of ResNet-family architectures for CT-based stroke classification, focusing on architectural trade-offs and evaluation rigor.
Machine learning–based anomaly detection for web application firewalls, integrating statistical learning and adaptive rule generation.
Hybrid ANN and meta-heuristic optimization approach for microstrip antenna design.
Deep learning–based image optimization pipelines designed for CDN environments, focusing on performance, bandwidth efficiency, and perceptual quality.
Designed and managed adaptive caching strategies driven by content popularity and access patterns, integrating data analysis with large-scale delivery systems.
Developed intelligent auto-scaling and tuning mechanisms for distributed infrastructure, leveraging predictive modeling and adaptive optimization.
Applied AI-driven optimization techniques to antenna geometry design, combining electromagnetic simulation with meta-heuristic search.
Designed and implemented an autonomous robotic boat platform, integrating control systems, sensor fusion, and embedded processing.
Developed an autonomous delta robot system using embedded vision, real-time control, and low-cost single-board computing.
Implemented closed-loop motor speed control on ARM-based platforms with serial and Bluetooth communication for real-time monitoring and control.
My methodology emphasizes reproducibility, comparative evaluation, and system-level thinking. I approach research problems by combining theoretical grounding with applied experimentation, ensuring that developed solutions remain transferable across domains—from medical imaging to distributed infrastructure and embedded systems.