Open to opportunities

Konstantin Sviblov

@konstantinsviblov

Message

Data scientist with 5+ years in NLP, LLMs, RAG, and MLOps.

Russian Federation

Message

What I'm looking for

I want to build production-grade NLP/LLM systems—especially search, matching, and RAG—where I can own infrastructure (pipelines, vector DBs, monitoring) and improve quality end-to-end with strong MLOps and reliability.

ML Engineer with 5+ years of experience across NLP, vector search, MLOps, and classic ML, focusing on production-grade search and matching systems and the infrastructure behind them.

Most recently I worked on product matching and search infrastructure for large marketplaces: vectorizing large product assortments into Qdrant, embedding research and comparison, cross-marketplace matching, and the supporting data pipelines (Airflow DAGs, ClickHouse analytics). I was responsible for the Qdrant vector database — optimization, production incident response (e.g. severe RAM overconsumption), and input into its planned migration to a dedicated server - and built the monitoring/alerting layer in Grafana (Kafka queue lag, parsing pipelines, data-quality alerts).

Previously, I built LLM/RAG applications end to end, worked with multiple vector databases (Qdrant, Milvus, CosmosDB) on Azure and AWS, and set up MLOps for a microservice-based project (Docker, MLflow, DVC). I also delivered a computer-vision project on IoT devices (Jetson Nano, AWS Greengrass/SageMaker/S3) and classic ML on tabular data (LightGBM, CatBoost).

I write clean, reproducible code (Git best practices, well-structured repositories) and am comfortable owning a system from data and experiments through to production and monitoring.

Before data science I worked as a design engineer, where I first picked up Python by automating routine work.

I can work as an independent contractor via my own legal entity (B2B) in Georgia. Open to remote roles within CET/GMT time zones

Experience

Work history, roles, and key accomplishments

ML Engineer

Wildberries

Dec 2024 - May 2026 (1 year 5 months)

- Designed and operated data pipelines using Airflow, managing ClickHouse-heavy analytics and setting up Grafana monitoring/alerting
- Developed, supported and troubleshooted the Qdrant vector database.
- Conducted embedding and LLM research to continuously enhance cross-marketplace matching and search accuracy.

Qdrant SQL Grafana Python Clickhouse NLP LLM

ML Engineer, Search & Match

Wildberries

Dec 2024 - May 2026 (1 year 5 months)

Built and operated daily Airflow pipelines and ClickHouse analytics with Grafana-based monitoring and data-quality alerts. Owned Qdrant for 300M-scale collections, handling performance optimization and incident response for RAM overconsumption and contributing to its migration to dedicated servers; delivered vectorized product matching and LLM/embedding experiments to improve search quality.

Python Airflow Clickhouse Qdrant Vector Search Grafana Monitoring And Alerting Docker Embeddings Incident Response

ML Engineer

WaveAccess

Jan 2024 - Aug 2024 (7 months)

- Developed a system for recommending correct MedDRA terms using both open-source and proprietary LLMs, various vectorizers, prompt engineering, and infrastructure tasks. Worked with Snowflake databases.
- Conducted research on the application using transformer model for predicting a specific disease in patients based on sequences of ICD codes. Worked with Snowflake databases

LLM Lang Chain Python

ML Engineer, LLM & RAG

WaveAccess

Jan 2024 - Aug 2024 (7 months)

Developed a system to recommend correct MedDRA terms using open-source and proprietary LLMs with RAG, prompt engineering, and vectorizer infrastructure, working with Snowflake. Researched transformer-based approaches for predicting disease from patient text sequences and ICD code inputs.

Python LLM RAG Prompt Engineering Vectorization Snowflake MedDRA Transformers Embeddings Experimentation

ML Engineer

EPAM

Apr 2021 - Dec 2023 (2 years 8 months)

A lot of projects in NLP, LLMs, IoT, CV, MLOps, Clouds.

NLP LLM RAG

ML Engineer, LLM & MLOps

EPAM

Apr 2021 - Dec 2023 (2 years 8 months)

Delivered multiple LLM-based products (AI math tutor, presentation summarization, support assistant) using RAG and document similarity search across vector databases and Azure services. Built an internal job-position matching and employment-status prediction system for 50,000+ employees (LightGBM to CatBoost), implemented MLOps with Argo CI/CD, MLflow and DVC, and developed an AWS-based fraud dete

Python LLM RAG Azure Vector Databases CatBoost MLFlow DVC AWS

Education

Degrees, certifications, and relevant coursework

SMTU

Specialist (Masters), Computers, complexes, systems and networks

2008 - 2013

Saint Petersburg State Marine Technical University

Specialist (Master's), Computers, complexes, systems and networks

2008 - 2013

Activities and societies: Projects included ML for AutoCAD drawings, a camera-based filament thickness measurement device (OpenCV/Python UI), and LoRa-based fine-tuning for an open-source LLM using personal text data.

Completed a Specialist (Master-level) program in Computers, complexes, systems and networks from 2008 to 2013.