Deep Reinforcement Learning for Donkey Kong
Applied DQN, PPO, and A2C to train agents in the classic Donkey Kong environment, with experiments and evaluations targeting learning stability and gameplay efficiency.
I am a Ph.D. student in the School of Electrical, Computer and Energy Engineering at Arizona State University and a member of the Performance and Resource Optimization in Networks (PROTON) Lab, advised by Dr. Eirini Eleni Tsiropoulou. I received my B.Sc. in Computer Science from Nazarbayev University in June 2025. Previously, I worked as a Research Assistant at Nazarbayev University on Federated Learning for medical IoT. I also worked on Computer Vision and Formal Verification of Distributed Systems.
Education
-
PhD in Computer Engineering
Arizona State University — School of Electrical, Computer and Energy Engineering
- Graduate Research Assistant in the Performance and Resource Optimization in Networks (PROTON) Lab
- Advised by Dr. Eirini Eleni Tsiropoulou
-
BS in Computer Science
Nazarbayev University
- cGPA of 3.4/4.0, top 12% of the cohort
- Dean's List, Spring 2023 — GPA 3.94
Research
My research focuses on optimizing distributed computing systems using game theory and reinforcement learning, with an emphasis on theoretical analysis and field validation.
Selected work
-
Federated Learning Framework for Human Activity and Anomaly Detection in IoMT
Research Assistant
- Collected and processed simulated vitals data in collaboration with NUSOM, populated with synthetic activity labels
- Implemented an updated federated learning framework for human activity recognition
-
AI-YM — Kazakh Sign Language
Research Assistant
- Implemented learning and quiz modules alongside a sign recognition solution for Kazakh Sign Language
-
Formal Verification of Repository Synchronization and Merge Consistency in a Distributed Version Control System
Formal Methods course research
- Modeled a repository management protocol in Promela and verified it with SPIN
- Extended the model with branches and basic conflict resolution
-
Multi-Scale Attention Network for Wheat Disease Classification
Computer Vision course research
- Developed MSANet with an EfficientNet backbone, multi-scale feature fusion, attention modules, and a classification head
Projects
Natural Disaster Tweet Classification
Compared classical models (Logistic Regression, SVM, Random Forests) against CNNs, LSTMs, and Transformers, surfacing strengths and failure modes for real-time disaster response.
Oil Spill Detection from SAR Imagery
Implemented U-Net with a ResNet18 backbone alongside YOLOv8 for marine oil spill detection, addressing confusion with look-alike phenomena.
Generating Rock Music from Spectrogram Images
Fine-tuned Stable Diffusion on mel spectrograms of rock clips from GTZAN, then reconstructed the generated images into audio with Griffin-Lim.
eGov Document Delivery Platform
Integrated the Public Service Centers information system, SMS Gateway 1414, Mobile Citizens Database, and the State Database of Physical Persons into a document delivery service.
Skills
- Programming Languages
- Frameworks & Libraries
- Tools