Looking for a job

Vadim Sharavin

@vadimsharavin

Message

I build reliable data pipelines and platforms for banking, fintech, and transportation.

Russian Federation

Message

I've built and automated data pipelines and platforms for Alfa-Bank, Sberbank, JSC NIIAS, and Russian Railways, working across banking, fintech, and transportation.

At Alfa-Bank, I developed a Python, watchdog, and Apache Airflow file orchestrator with four DAGs for all 10 regulatory reports. I also reverse-engineered Greenplum, ClickHouse, and MS SQL sources, prepared CDC integration requirements for a Debezium → Kafka → Flink architecture, and built four SAP OData parsers for the DWH team.

At Sberbank, I optimized PySpark ETL on YARN by addressing data skew and redesigned customer identification logic, eliminating 180,000 duplicate or invalid records. I also built an incremental SCD Type 2 pipeline and maintained Airflow DAGs supporting a customer-verification product for more than 4 million customers.

Earlier, I built a PostgreSQL DWH and Python ETL pipelines at JSC NIIAS, eliminating manual Excel processing and reducing legacy dashboard load times from 45–60 seconds to 8–10 seconds. At Russian Railways, I automated production file deployment and PDF data extraction, turning hours of schedule preparation into 3–10 minutes.

Experience

Work history, roles, and key accomplishments

Data Engineer

Alfa-Bank

Dec 2025 - Aug 2026 (8 months)

- Developed and tested a file orchestrator using Python, watchdog, and Apache Airflow: 4 DAGs automated file-processing pipelines for all 10 regulatory reports; implemented integrity checks, retries, quarantine, logging, and email notifications
- Reverse-engineered data sources and prepared them for integration: analyzed DDL and data structures in Greenplum, ClickHouse, and MS SQL, designed STT ma

Data Engineer

Sberbank

May 2025 - Dec 2025 (7 months)

- Optimized PySpark pipeline performance on a YARN cluster, resolving data skew with broadcast joins, repartitioning, and salting, reducing partition imbalance and accelerating ETL job execution
- Redesigned customer identification logic at the join level in a PySpark pipeline (migrated from a shared tax ID to a unique customer key), eliminating record loss during aggregation and removing 180,000

PySpark Batch Processing Data Analysis Big Data SQL Hive ETL Hadoop Apache Spark Banking A B Testing Query Optimization Qlik Sense Jenkins Airflow Data Quality Data Governance

Data Engineer

JSC NIIAS

Sep 2023 - May 2025 (1 year 8 months)

- Designed and built a PostgreSQL DWH from scratch (Staging and Data Mart layers), replacing local CSV/XLSX file storage with a centralized data source for Power BI and Superset
- Built ETL pipelines in Python (pandas, SQLAlchemy) for extracting, validating (reference data checks, range validation), and loading data from 5+ sources into PostgreSQL, fully eliminating manual processing in Excel
- Im

Git Power BI Data Marts Data Analysis SQL ETL Apache Superset SQLAlchemy Data Modeling Query Optimization NumPy Jira Power Query Pandas DAX Python Polars

Engineer

Russian Railways

Sep 2021 - Sep 2023 (2 years)

- Built a bash automation script to deploy files across multiple production servers, replacing a manual ~1-hour process with a few minutes and eliminating human error from the workflow
- Automated PDF data extraction and processing using Python (pdfplumber, pandas), converting unstructured documents into structured datasets and cutting schedule preparation time from several hours to 3–10 minutes

Oracle Database Python BASH Pandas SQL Automation DWH