← Back to list
senior
Registration: 27.03.2025

Dmitriy Ulybin

Specialization: Machine Learning Engineer
— Senior Machine Learning Engineer with over 5 years of experience in developing and deploying AI solutions. — Specialized in NLP, search, and recommendation systems, with expertise in RAG, LLMs, and vector search. — Experienced in building end-to-end ML pipelines, optimizing ranking algorithms, and integrating AI-driven solutions into production.
Machine Learning NLP AI LLM
— Senior Machine Learning Engineer with over 5 years of experience in developing and deploying AI solutions. — Specialized in NLP, search, and recommendation systems, with expertise in RAG, LLMs, and vector search. — Experienced in building end-to-end ML pipelines, optimizing ranking algorithms, and integrating AI-driven solutions into production.
Machine Learning NLP AI LLM

Portfolio

Xena Exchange

● Uplift Modeling & ETL Automation for Crypto Exchange. Uplift Modeling: ● Developed a classification model (CatBoost) to predict the impact of ad campaigns on user conversions. ● Built a data processing pipeline (Python, Pandas) and applied feature engineering, improving F1-score to 0.89. ● Impact: Increased conversion rates by 22% for users receiving targeted offers (account top-ups). ETL Automation: ● Designed a data collection system aggregating information from 9 sources (REST API + asynchronous parsing via BeautifulSoup and Selenium). ● Orchestrated ETL workflows in Airflow with DAGs for hourly data updates. ● Integrated structured data into PostgreSQL and visualized analytics with Plotly. ● Impact: Reduced data retrieval time for analysts by 2–3x.

IT Solutions

● Developed ML solutions for industry clients, including: - Oil & Gas: Failure prediction for industrial equipment. - Transportation: Forecasting the month when railcars require maintenance. - Aviation: Ticket ranking model to optimize customer choices. ● Worked on end-to-end ML pipelines, from data preprocessing to model deployment. ● Grew from Intern to Full-Time Specialist.

MTS

● AI-powered Content Navigator (NLP, RecSys) for Movies, Books, Places, and Events. ● Built a RAG-based search system using Vespa vector DB and advanced embedding models, significantly enhancing retrieval quality. ● Improved search relevance through: - Intent classification for accurate query categorization and filtering. - Paraphrase generation via LLM to enhance semantic search and retrieval. - Entity recognition (NER) for precise matching of key information. ● Alignment of LLM and fine-tuning transformers to adapt language models for specific tasks and improve performance. ● Optimized geosearch to enhance search accuracy based on geographical context. ● Increased NDCG and MAP metrics, refining the relevance of the recommended content. ● Designed a microservices architecture with a FastAPI pipeline, orchestrated RAG workflows, and deployed scalable solutions via Docker.

Skills

Machine Learning
AI
NLP
LLM
RAG
Search
Data Management

Work experience

Senior Machine Learning Engineer
since 04.2024 - Till the present day |MTS
Python, Vespa, PyTorch, Transformers, VLLM, Text Embeddings Inference, FastAPI, Docker, Kubernetes
● AI-powered Content Navigator (NLP, RecSys) for Movies, Books, Places, and Events. ● Built a RAG-based search system using Vespa vector DB and advanced embedding models, significantly enhancing retrieval quality. ● Improved search relevance through: - Intent classification for accurate query categorization and filtering. - Paraphrase generation via LLM to enhance semantic search and retrieval. - Entity recognition (NER) for precise matching of key information. ● Alignment of LLM and fine-tuning transformers to adapt language models for specific tasks and improve performance. ● Optimized geosearch to enhance search accuracy based on geographical context. ● Increased NDCG and MAP metrics, refining the relevance of the recommended content. ● Designed a microservices architecture with a FastAPI pipeline, orchestrated RAG workflows, and deployed scalable solutions via Docker.
Machine Learning Engineer
12.2022 - 04.2024 |MTS
Python, PyTorch Lightning, Transformers, ONNX, FastAPI, Docker, Triton Inference Server
● AI-powered Vulnerability Filtering Service (ASOC) for Source Code Security. ● Developed an ensemble model combining: - Static code analyzer features. - Code embeddings (fine-tuned CodeBERT on a vulnerability dataset). ● Optimized inference with Triton Inference Server using ONNX conversion, dynamic batching, and FP16 quantization. ● Integrated into CI/CD pipelines, reducing false positives by 70%, significantly decreasing AppSec engineers’ workload.
Data Scientist
01.2021 - 11.2022 |Xena Exchange
Python, CatBoost, Scikit-learn, Airflow, PostgreSQL, Docker, Selenium, Plotly
● Uplift Modeling & ETL Automation for Crypto Exchange. Uplift Modeling: ● Developed a classification model (CatBoost) to predict the impact of ad campaigns on user conversions. ● Built a data processing pipeline (Python, Pandas) and applied feature engineering, improving F1-score to 0.89. ● Impact: Increased conversion rates by 22% for users receiving targeted offers (account top-ups). ETL Automation: ● Designed a data collection system aggregating information from 9 sources (REST API + asynchronous parsing via BeautifulSoup and Selenium). ● Orchestrated ETL workflows in Airflow with DAGs for hourly data updates. ● Integrated structured data into PostgreSQL and visualized analytics with Plotly. ● Impact: Reduced data retrieval time for analysts by 2–3x.
Data Scientist
02.2020 - 12.2020 |IT Solutions
ML, Scikit-learn, XGBoost, CatBoost, Python, Pandas, NumPy
● Developed ML solutions for industry clients, including: - Oil & Gas: Failure prediction for industrial equipment. - Transportation: Forecasting the month when railcars require maintenance. - Aviation: Ticket ranking model to optimize customer choices. ● Worked on end-to-end ML pipelines, from data preprocessing to model deployment. ● Grew from Intern to Full-Time Specialist.

Educational background

Data analysis
Till 2022
NRU ITMO

Languages

EnglishProficientRussianNative
Machine LearningNLPAI