I am a Systems & AI Researcher at Nazarbayev University bridging complex mathematical theory with highly optimized, low-level execution. I specialize in C++ high-performance computing, WebAssembly, and hardware-aware deep learning optimizations on constrained silicon (Apple MLX).
- NogaiLLM: Curing Catastrophic Forgetting in Zero-Resource Turkic NLP | Under Review: ACM TALLIP
- Engineered a memory-safe Apple MLX pipeline to execute Parameter-Efficient Continuous Pre-Training (CPT) and Supervised Fine-Tuning (SFT) natively on consumer hardware.
- 9Q Engine: Strongly Bounding a 10^25 Game Tree | Under Review: IEEE ToG
- Built a 1-billion game C++ Minimax engine with 16-byte aligned Transposition Tables, mathematically bounding the state-space and cross-compiling to WebAssembly (186K+ NPS).
- Upcoming: Strongly solving "Bestemshe" via Retrograde Analysis (15GB Tablebase) and introducing RCL-ZERO (Retrograde Curriculum Learning for DAG-game Neural Networks).
Languages: C++, Python, SQL AI & Hardware Optimization: Apple MLX, PyTorch, QLoRA, Tensor Casting, PEFT Systems: WebAssembly (Emscripten), Multi-threading, CMake, GitHub Actions CI/CD, GTest