Supporting organization building foundation of Yandex Cloud (public cloud like AWS):- network virtualization and network services: VPC, NLB, ALB, DNS, etc- compute virtualization (EC2-like)- network block storage (EBS-like)- public APIs: cli/api/sdk/terraform- internal developer experience

At Yandex Go (internationally as Yango), I was leading the tech strategy for a $10bn+ annual GMV ride-hailing (Uber-like) service, managing 8 million+ daily rides via an 800+ microservice backend.Have led improvement of overall business reliability (uptime) from 99.9% to 99.99% by: setting strategy & vision, integrating SRE practices, building a reliability infrastructure, hiring & training a reliability team, redesigning the system (e.g., start transitioning to DOMA and pubsub), enhancing observability, establishing incident management, and more.

Supporting 13 product and 2 infra teams in backend of Yandex Go (Yango) and Scooters. Leading software integration and due diligence in M&A of Wind https://techcrunch.com/2021/10/12/russian-tech-giant-yandex-expands-israeli-footprint-with-acquisition-of-winds-tel-aviv-operations/

Scaling product backend engineering from 3 to ~10 teams. Supporting teams to launch from 0 to 1: super app, A/B testing platform and company-wide cashback.Leading performance optimization of Yandex Go application: -2.4s p50 main screen loading time.

Scaling team from 4 to 12 backend developers. Examples of launched features:1. passenger location for drivers - http://actualapple.com/yandex-taxi-allowed-users-to-show-their-location-to-drivers/2. driver fatigue monitoring - https://eng.crownmedia.su/2018/05/28/yandex-taxi-introduces-a-system-for-monitoring-driver-fatigue/

Supporting backend (go) and frontend (react, redux, typescript, javascript) team at BeepCar.BeepCar is carpooling platform and it was acquired by BlaBlaCarhttps://venturebeat.com/2018/08/06/blablacar-acquires-russian-carpooling-platform-beepcar-from-mail-ru/

Launching backend for startup BeepCar from the 0 to 1 and supporting backend team. Using OpenStreetMap, ElasticSearch and mysql at core.

1. Reducing number of frontend machines by 30% (saved 110 machines) by optimizing code2. Decreasing message metadata database timeout failures rate by 2 times3. Raising the quality bar in a team: introducing unit tests, functional tests, fuzzy testing, sanitizer and valgrind testing, static analysis.

1. Launching from 0 to 1 new emails storage on scale of 10 petabytes and 4000 servers. The launch was performed without any errors on production. Using C as a programming language.2. Optimizing 3x latency of service responsible for attached to email files.

Supporting highly scalable backend servers with 11k RPS: debugging core-dumps, fixing bugs, adding features. Supporting nginx mail proxy module.Decreasing 10x failure rate on primary API (messages list) by switching to new database.