ML Engineer & AI Researcher | Master's Student in AI @ Chung-Ang University, Seoul
Building efficient AI systems with focus on vision-language models, curriculum learning, and feature selection.
VLM Compression – Optimization techniques for vision-language models Benchmarked quantization (INT8, INT4) and structural pruning (MLP, KVQ) on BLIP-2 and LLaVA, achieving 47% latency reduction
Curriculum Learning Fine-Tuning – Mathematical reasoning with curriculum strategies Fine-tuned SmolLM2 and Phi-3 on GSM8K with complexity-based curriculum, achieving 1.37% accuracy improvement
BAGen – Automated content generation for children's music End-to-end pipeline using LLMs and diffusion models to generate synchronized lyrics and animated backgrounds
Feature Selection Library – Information-theoretic methods Implemented 30+ unsupervised algorithms with 25 datasets for Master's thesis research
Journals:
- ChainImputer: A Neural Network-Based Iterative Imputation Method Using Cumulative Features – Symmetry (2025) | GitHub
- EP-REx: Evidence-Preserving Receptive-Field Expansion for Efficient Crack Segmentation – Symmetry (2025)
Conferences:
- Diffusion Model-Based Generative Pipeline for Children Song Video – IEEE ICCE (2025)
C++ Software Developer @ YADRO (May 2023 - Feb 2024) Real-time signal processing for 4G base stations | AVX SIMD optimization | Telecommunication systems
Languages: Python, C++, TypeScript ML/AI: PyTorch, Transformers, Diffusion Models, Quantization, Pruning Tools: Git, Docker, HuggingFace, MongoDB
Portfolio: khrtim.github.io LinkedIn: linkedin.com/in/timur-khairulov-b09791250
Open to ML Engineer positions and research collaborations