Alexander Kolesen

Self-employed · remote

Developing a cloud orchestration product for ETL and data-processing workloads, with end-to-end responsibility for architecture and implementation.

• Design and implement AWS infrastructure for ETL and data-processing workloads using Step Functions, ECS, Batch, and S3.
• Manage the full infrastructure lifecycle with Terraform, ensuring reliable and observable operation of production services.

Stack: Terraform, Python, PostgreSQL, AWS (Step Functions, ECS, Batch, Lambda, S3, RDS, CloudWatch)

Online Services, People Can Fly · ~600 people · Warsaw, PL (remote)

People Can Fly is a global video game development company.

• Built and launched the foundation for a self-hosted online game platform serving 600+ developers and supporting multiple new titles.
• Built the content pipeline for the staging environment for content developers.
• Partnered with internal stakeholders and external game-hosting providers (Unity Multiplay, AWS GameLift) to gather requirements, evaluate options, define the target architecture, and coordinate implementation.
• Owned the platform's production readiness: deployment processes, observability, reliability requirements, and coordination of operational responsibilities.
• Built the CI/CD pipeline for the game platform.

Stack: AWS, Terraform, Java, Kotlin, GitHub Actions, GameLift, Multiplay

ActVPN and wscp, Actmobile · ~10 people · San Francisco, US (part-time, remote)

Actmobile builds VPN products. Worked on ActVPN, a VPN written in C, and wscp, a maximum-throughput network copying tool.

• Integrated LwIP, a userspace TCP/IP stack, into ActVPN.
• Tracked down memory leaks and stabilized ActVPN.
• Implemented a dual-channel model (UDP for data, TCP for retransmitting lost fragments) that outperformed standard TCP by 2–10× in bandwidth utilization under certain network conditions.

Stack: Linux, C, LwIP, OpenSSH, rsync, TCP, UDP

Palta Data Platform · ~10 people · Limassol, CY (part-time, remote)

Palta is a startup incubator for health and well-being applications; its data platform replaced third-party data collection services with an internal one.

• Designed and implemented a cost-effective, multi-tenant data pipeline (API Gateway → Kinesis Firehose → S3 → SQS → Lambda → Snowflake) at one-tenth the cost of Amplitude on the same data volume.
• Designed and implemented a reliable HTTP callbacks system.
• Designed infrastructure for a payments system to support rapid load scaling.

Stack: Docker, PostgreSQL, Apache Kafka, Snowflake, Terraform, Atlantis, Grafana, GitHub Actions, AWS (API Gateway, Lambda, S3, SQS, Kinesis, DynamoDB, RDS, CloudWatch)

ICE, Denmark's Ministry of Taxation (Flexiana) · 500–1000 people · Copenhagen, DK (remote)

A public system covering the full property taxation process in Denmark, built in Clojure/ClojureScript on AWS.

• Maintained and extended several components to eliminate performance bottlenecks.
• Recognized and bridged a gap between developers and infrastructure engineers, significantly increasing the velocity of infrastructure-related delivery.
• Sped up the CD pipeline 4–10× by introducing more granular deployments.

Stack: Clojure, Docker, PostgreSQL, Jenkins, Terraform, Atlantis, GitHub Actions, AWS (Batch, API Gateway, EC2, ECS, Lambda, S3, SQS, CloudWatch, RDS)

Wanna Kicks · ~20 people · Minsk, BY (part-time, remote)

Virtual try-on application that lets users see how sneakers would look on their feet before buying.

• Implemented a CRUD backend to manage the model catalog.
• Distributed the model catalog as encrypted bundles via CDN, in a multi-versioned and multi-tenant fashion.
• Implemented statistics collection and SQL-based analysis using AWS Kinesis Firehose and AWS Athena.

Stack: Python, Chalice, Docker, Terraform, AWS (API Gateway, Lambda, S3, SQS, DynamoDB, CloudFront)

Juno · 100–200 people · Minsk, BY — New York, US (onsite in Minsk)

Juno was a ride-sharing service operating in New York, acquired by Gett in 2017.

• Scaled from a fresh team to 40K daily trips in under one year in the highly competitive NYC market, serving millions of users.
• Established continuous delivery of 100+ backend microservices, running several times a day and sometimes exceeding 600 deploys daily.
• Worked on capacity planning and fault tolerance, isolating parts of the computing cluster into separate shards.
• Led a team of 5 infrastructure engineers serving 40 core backend engineers and 15 QA engineers.
• Provided developers with tooling to inspect production logs and metrics and to maintain their own alert sets.

Stack: Linux, Clojure, Golang, Python, MySQL/InnoDB, PostgreSQL, MongoDB, Docker, GoCD, nginx, Elasticsearch, Kibana, Prometheus, Grafana, AWS (EC2, ECS, RDS, S3, ElastiCache, Route 53)

Kontur · Dyn (via EPAM) · Iron.io · Wargaming.net

• Kontur, Inc. (GIS service provider) — built backend infrastructure for Kontur Platform Services (Kubernetes, Java, nginx).
• Dyn Inc. (DNS, via EPAM, acquired by Oracle) — designed an API bridging legacy and new DNS systems; built Docker + Chef + Jenkins pipelines.
• Iron.io (cloud platform) — migrated MongoDB clusters, automated monitoring, extended deployment infrastructure.
• Wargaming.net (gaming) — built deployment tooling and the infrastructure team; scaled web systems to 1M+ concurrent players.