Machine Learning Engineer
Presidential Affairs Administration of Kazakhstan, Engineering & Technical Center
May 2026 — Jun 2026Built a request router for specialized AI models and evaluated internal ML and LLM systems to identify routing failures and unsafe edge cases.
• Designed an AI model-routing system that classified incoming requests and selected specialized models based on task type and capability, enabling more efficient use of multiple AI systems. • Evaluated internal ML and LLM systems for failure cases, routing weaknesses, and unsafe edge cases; proposed improvements to model selection and system reliability.
Machine Learning Engineer
Uyren AI
May 2025 — Sep 2025Built a lecture-grounded LLM assistant with video timestamps and an end-to-end RAG pipeline, reducing token use by about 40%.
• Built an end-to-end LLM educational assistant using GPT-4o and Claude APIs, grounding answers in lecture content and linking users directly to relevant video segments and timestamps. • Designed the full RAG pipeline, including ingestion, chunking, embeddings, ChromaDB retrieval, context construction, generation, and evaluation; reduced token usage by about 40% through retrieval and context optimization.
Machine Learning Engineer
Traffic Analytics Solutions
Jan 2024 — Apr 2025Deployed a YOLO traffic analytics system that automated video review and reached up to 98% accuracy on tasks, including in night and snow conditions.
• Developed and deployed a YOLO-based traffic analytics system for vehicle counting, tracking, lane changes, turning directions, and intersection analysis. The system became part of the company’s operational workflow. • Automated work previously performed through manual video review, substantially reducing staff time required for traffic counting and analysis. Improved performance in night and snow conditions, reaching up to 98% accuracy on deployed tasks.
Research Assistant, AI Evaluation
ISSAI, Nazarbayev University
ISSAI is Nazarbayev University's Institute of Smart Systems and Artificial Intelligence, a research institute focused on AI and related technologies.
Jun 2026 — Jul 2026Won one of six funded positions from 1,200+ applicants and built evaluation pipelines for research on Vision-Language Models as automated judges.
• Awarded one of only 6 funded research positions from 1,200+ applicants; researched the reliability of Vision-Language Models used as automated judges. • Built reproducible multi-model API evaluation pipelines and led development of the automated experimental workflow for an EACL 2027 research submission, a CORE A-ranked AI/NLP conference.
