About me
I'm a Data Science student who treats every project as a production system.
I don't just train models — I architect pipelines, deploy APIs, build UIs, and ship things people can actually use.
- Currently at EY Academy of Business · ex-Backend dev at Agency of RK for Financial Market Regulation
- Hackathon competitor — always the solo technical contributor on the team
- Interested in: explainability (SHAP), multi-agent systems, RAG, local LLMs
- Languages: Kazakh (native) · Russian (C2) · English (B2)
Currently learning: LLM fine-tuning · computer vision at scale · MLOps
AI / Machine Learning
Backend
Frontend
DevOps & Tools
| Project | Stack | Description |
|---|---|---|
| DTC Central Asia | Django · PostgreSQL · Docker · MinIO | Commercial e-commerce platform for furniture hardware — built, deployed, maintained |
| Qazaq Genius | Django · React · Docker · MinIO · PostgreSQL | Platform for translating Kazakh songs preserving cultural context and emotion |
| Project | Stack | Description |
|---|---|---|
| Saga Platform | FastAPI · LangChain · MCP · Docker | Multi-agent AI for meal planning. CV food analysis + RAG + auto cart via VkusVill MCP |
| AI Lawyer | Saiga-2 · Streamlit · Vector DB | Legal assistant for Kazakhstan law — RAG kills hallucinations, semantic search over legal code |
| Car Damage Classifier | PyTorch · FastAPI · Docker | CNN for vehicle damage detection. Full pipeline: augmentation → training → REST API. Acc — 95% |
| LLM Book Translator | Python · Aiogram · BeautifulSoup4 | Book translation bot with dynamic glossary to keep context across large volumes |
| Project | Stack | Result |
|---|---|---|
| Churn Prediction | CatBoost · SHAP · Pandas | 3-class model on 28M+ rows, cascading architecture — HackNU 2026 finalist |
| PathoAI | UNet++ · MAnet · EfficientNet-B4 | Medical biopsy segmentation, val IoU ~0.83 — AI in Healthcare 2026 |
| QazGeoAI | CatBoost · RuBERT · ChromaDB | Geospatial investment analysis for Kazakhstan — Decentrathon 5.0 |
Backend Developer — Agency of RK for Financial Market Regulation · Sep 2025 – Feb 2026
Fault-tolerant backend on Django, full cycle from DB schema design to deployment.
Digital Marketer — EY Academy of Business · Feb 2026 – present
Brand identity compliance, marketing materials, digital asset management.
