Operative card
- Class
- Data engineer
- Employer
- Welltory, lakehouse on AWS
- Exp
- 9 years, since 2017
- Node
- Montevideo, UY · remote
- Status
- Open to consulting
Dmitrii Cherkashin
Data engineer · snowboarder
I build data platforms that keep working when the data grows 80×. Nine years, six companies, one lakehouse built from scratch.
Survived an 80× data surge
Contract log
Experience, newest first, and what changed because I was there.
-
2025–now
WelltoryData engineer
Built the company’s lakehouse on AWS from scratch. Took over an abandoned open-source DynamoDB connector and cut its write load 5.7×. The CDC pipeline held through an 80× data surge.
-
2024
US music publisherFreelance contract
Replicated an IBM AS/400 system into ClickHouse, with reconciliation and recovery tooling.
-
2023–2024
AnabarSenior data engineer
Data processing down from 17 h to 7 h, p90 queries from 20 s to under 1 s, ClickHouse storage −40%.
-
2021–2022
Agro.clubLead data engineer
Hired the data warehouse team from zero. Business reports went from 1.5 hours to 2 minutes.
-
2020–2021
CitymobilSenior data engineer
Built CDC from MySQL through Kafka into Exasol, plus one-click replication for new tables.
-
2018–2020
Mail.ru GroupSenior system analyst, DWH
Designed warehouse models for Delivery Club and Citymobil. Task-to-report time fell from 3 weeks to 1.
-
2017–2018
Tinkoff BankSystem analyst, DWH
Integrated the bank with the national payment card system and designed Data Vault models for its warehouse.
Installed cyberware
The stack I run in production right now, plus what is still in the bloodstream.
- Lakehouse
- Apache Iceberg, AWS S3, Glue, Athena, Lake Formation
- Streaming
- Kafka, Kafka Connect, Debezium, CDC
- Orchestration
- Airflow, dbt, Spark
- Governance
- OpenMetadata, data quality, GDPR pipelines
- Warehouses
- ClickHouse, Trino, Greenplum, Vertica, Exasol, Data Vault
- Languages
- Python, SQL, Scala
- Infra
- Docker, Kubernetes, Terraform, Grafana
Side gigs
Built after hours, used every day.
-
Vox
Meeting recorder for macOS. Local transcription and speaker diarization: the audio never leaves the laptop.
Swift 6 · SwiftUI · Parakeet · pyannote
-
dima-lab
Home server running my self-hosted stack: monitoring, smart home, a Git server and an AI agent, all behind a private mesh network.
Docker · Home Assistant · Grafana · Caddy · Tailscale
-
dimstunt-plugins
Plugins and skills for Claude Code: a knowledge wiki maintained by agents, homelab ops, pre-merge review gates.
Claude Code · Python · MCP
Repos are private. Happy to walk through any of them on a call.