Yermukhammed Temerkulov

@temyer

AI/ML Engineer

I am an ML and NLP Engineer with a foundation in full-stack development and MLOps.

About

My core interests lie in Generative AI, specifically designing multi-agent LLM architectures, building robust RAG pipelines, and fine-tuning models for real-world applications. I am passionate about owning the end-to-end ML lifecycle, from predictive modeling to scalable deployment, which I am currently deepening through my M.Sc. in Machine Learning.

Career

Experience: 6 years

Experience

NLP Engineer

Sber

Russian banking and technology group.

Apr 2025 — Present

Built LLM-based multi-agent and RAG systems; the internal BI system reduced data-lookup time by about 60%, and the external Q&A system achieved 85%+ answer relevance.

• Architected and deployed a multi-agent system for the internal BI platform “Navigator” serving 5,000+ employees, implementing a supervisor architecture with specialized agents for dashboard, widget, and document retrieval powered by GigaChat; reduced average data-lookup time by ~60%. • Designed and built a RAG pipeline end-to-end as a solo developer—FastAPI backend, OpenSearch vector store, and retrieval-augmented generation layer (E5 embeddings + GigaChat)—serving as a document Q&A system for external clients with 85%+ answer relevance.

ML Engineer; SWE / MLOps Engineer; Frontend Developer

MAXIMUM Education

Jul 2019 — Jan 2025

Progressed from frontend development through SWE/MLOps into ML engineering, building analytics and serving infrastructure, computer vision and LLM services, and a lead-scoring model that doubled conversion from 2% to 4%.

• Developed and deployed an LLM-powered service for auto-generating support responses, fine-tuned on 10K+ proprietary tickets; reduced average first-response time by ~90% and increased ticket throughput by 30%. • Built a lead-scoring model (gradient boosting + feature engineering on 50K+ records) that doubled conversion rates from 2% to 4% for a key product line; implemented a daily retraining pipeline with automated data delivery to 15+ sales managers via internal dashboards.

Frontend Developer

Cheq

Apr 2018 — Jul 2019

Developed Vue.js and Nuxt.js web applications, integrated APIs, and added state management and end-to-end tests.

• Developed responsive web applications and SPAs using Vue.js, Nuxt.js, and Node.js; delivered pixel-perfect implementations from Figma designs. • Integrated RESTful APIs, implemented state management (Vuex), and wrote E2E tests to ensure production reliability.

Education

M.Sc., Machine Learning

Central University

2026 — 2028

B.Sc., Computer Systems & Information Technology

National Research Nuclear University MEPhI

2025

B.Sc., Physics

Université de Montpellier

2017

Selected projects

Telegram bot for university staff data collection

Automated university staff data collection, reducing operational overhead and manual processing time by ~70%.

Achievements

2nd Place, HSE × Sber ML Hackathon

2023

2nd Place, CU × Tutu Hackathon

2024

Skills

PythonMLAI