Dmitry Davydov

Staff Backend Engineer | Java/JVM | Distributed Systems | Platform & Reliability
Mexico | CST | Remote | linkedin.com/in/oroboros

Professional Summary

Staff backend engineer with 20+ years in production software, including 11 years at Yandex / Yandex Market. Primarily a hands-on Java/JVM engineer: built shared backend infrastructure, distributed data systems, billing and antifraud services, database and messaging integrations, observability libraries, overload protection, and live migrations. Recent work centered on a Java platform adopted by 350 production services; earlier work included large-scale analytics infrastructure and money-sensitive distributed processing. Strong in distributed systems, platform engineering, reliability, PostgreSQL, ZooKeeper/Curator, Kafka / Yandex Logbroker, performance, concurrency, and debugging difficult production failures.

Core Skills

Distributed Systems: ZooKeeper/Curator, high availability, service coordination, distributed queues, service discovery, live migrations, consistency and failure handling, load shedding

Java & Backend: Java, JVM internals, Spring Boot/Spring, multithreading and concurrency, HTTP APIs, protobuf/OpenAPI, jOOQ, application architecture, system design

Storage & Databases: PostgreSQL, Oracle, Redis, ClickHouse, Vertica, SQL, transactions, partitioning, query optimization, OLTP/OLAP

Distributed Data & Messaging: Kafka / Yandex Logbroker, Hadoop/HDFS/MapReduce, Yandex YT/YQL and dynamic tables, distributed storage, data pipelines, CDC

Cloud & Infrastructure: AWS - recent hands-on with ECS/Fargate, RDS, S3, SQS, CloudWatch, IAM; production background primarily on large-scale internal cloud/platform infrastructure. Linux, Docker, Terraform, OpenTelemetry, metrics, logging, monitoring, CI/CD

Languages: Java, Python, SQL, Bash; working Go and Kotlin

Professional Experience

Yandex / Yandex Market - Staff Engineer, Java Developer Platform | Nov 2022-Jun 2025

Java/Kotlin, Spring Boot, PostgreSQL, Redis, queues/messaging, OpenTelemetry, Docker, Terraform; large-scale internal AWS-like cloud with EC2-like VM and Fargate-like container infrastructure.

  • Built and evolved a Spring Boot-based platform used by 350 of roughly 600 Java services and about 2,000 developers, writing shared libraries and integrations for databases, queues, service discovery, auth/secrets, telemetry, deployment, and testing.
  • Implemented infrastructure drivers for database queues and messaging, core auth libraries, Docker packaging, and the migration from custom telemetry to OpenTelemetry; debugged difficult production integrations across the service fleet.
  • Designed and implemented adaptive CPU load shedding using jitter and leaky-bucket ideas; in favorable CPU-bound services it sustained up to roughly 1.5x more traffic without major response-time degradation and kept overloaded instances from collapsing under spikes.
  • Fixed or standardized failure-sensitive defaults including PostgreSQL multi-host failover timeouts, DNS caching, circuit-breaker behavior, deadline propagation, and JVM/runtime settings, turning recurring production lessons into fleet-wide protections.
  • Orchestrated fleet-wide rollouts across an internal AWS-like cloud running EC2-like VMs and Fargate-like container workloads, using early adopters, fleet metrics, criticality ordering, and fast rollback paths to ship cross-cutting changes safely to hundreds of services.

Yandex / Yandex Market - Senior to Staff Engineer, Analytics & Data Infrastructure | 2019-2022

Java, Spring Boot, ZooKeeper/Curator, YT/YQL, dynamic tables, PostgreSQL, Redis, ClickHouse, Kafka / Yandex Logbroker, Python.

  • Took technical ownership of a failing analytics platform serving about 150 analysts and 20 data engineers, with roughly 20 PB of data and 10K-30K cores of distributed compute; turned it into predictable infrastructure with explicit SLAs, monitoring, ownership, and capacity planning.
  • Built and operated a 60-node ClickHouse cluster and Java/Python orchestration that loaded roughly 20 TB of new or updated data from distributed storage each day into a stable serving layer.
  • Built a computation dependency-graph system for pipelines with hundreds of nodes, making hidden dependencies, bottlenecks, and late data visible; worked across ZooKeeper/Curator, YT, dynamic tables, PostgreSQL, Redis, and Kafka / Yandex Logbroker-backed systems.
  • Established protected resource pools, data-readiness SLAs, and capacity forecasting; worked with more than 50 source teams on reliable raw-data delivery, CDC adoption, schema changes, and data quality.
  • Worked on online antifraud data flow that ingested events from Kafka / Yandex Logbroker streams for low-latency evaluation, complementing the larger batch-processing path.
  • Promoted to Senior in 2020 and Staff in 2022 as the scope of hands-on architecture and implementation expanded across the wider analytics platform.

Yandex / Yandex Market - Middle Software Engineer, Billing Data & Antifraud | 2015-2019

Java, Spring, ZooKeeper/Curator, Hadoop/HDFS/MapReduce, Yandex YT/YQL and dynamic tables, PostgreSQL, Vertica, Kafka / Yandex Logbroker.

  • Migrated business-critical click billing and antifraud processing from Oracle to Hadoop and later to Yandex's in-house YT distributed platform, keeping live money flows available and preserving correctness during months-long dual-run migrations.
  • Owned multiple Java backends, APIs, ZooKeeper/Curator-based distributed components, PostgreSQL queries, and large-scale processing jobs with high-availability requirements over hundreds of terabytes of data.
  • Created online antifraud by moving suitable checks out of 40-120 minute batch pipelines into in-memory evaluation; the online filters caught roughly 80% of fraud traffic before billing.
  • Moved most remaining antifraud filters to partitioned PostgreSQL, cutting typical runtimes from roughly 40-120 minutes to 20-25 minutes and making the logic substantially easier to test and reason about.
  • Designed long-running migrations with old and new systems operating in parallel, compatibility checks around legacy behavior, gradual cutovers, and production verification so business-critical services could move without planned downtime.
  • Prototyped supporting data infrastructure including Vertica loading, incremental extraction from relational databases, data-quality checks, and reliable staged uploads for large source tables.

Yandex / Yandex Market - Junior Software Engineer, Billing Core | 2014-2015

  • Maintained money-sensitive Java services and APIs backed by a multi-terabyte Oracle database with thousands of tables and hundreds of large legacy SQL queries.
  • Built a new highly available backend service for partner delivery configuration while continuing day-to-day work on the legacy billing system; also introduced a test kit for a large 14-year-old service.
  • Ran the service across roughly 70 small nodes with immediate configuration visibility requirements and learned first-hand how health checks, database dependencies, and load-balancer timeouts interact in production.

CNII EISU - Java Security / Infrastructure Engineer | 2012-2014

  • Integrated identity and permissions with Java Security Manager, Java EE, OSGi, JBoss, APIs, databases, and application infrastructure.
  • Built a Javassist bytecode patcher that enforced annotation-based access control, requiring deep work with JVM, bytecode, framework, and application-server internals.
  • Created demo applications and integration infrastructure so identity and permissions could propagate transparently through APIs, databases, web servers, application servers, and Java middleware.

dietaonline.ru - CTO / Sole Technical Owner | 2008-2012

  • Built and operated the full product stack for a health and fitness website reaching about 1M monthly visitors, including PHP/MySQL, Sphinx, nginx, memcached, Linux operations, backups, monitoring, and frontend work.
  • Continuously optimized SQL, caching, code, and infrastructure to run the service on a single modest server under tight hardware constraints.
  • At peak the site handled about 10K hourly sessions and roughly 20 HTTP requests per second, translating to around 200 database queries per second; query tuning and caching were recurring engineering work.
  • Owned Linux administration, monitoring, backups, deployment, incident response, and security with no separate operations team.

Planned Sabbatical

2025-2026: Planned career break after Yandex; traveled in Latin America and prepared to return to hands-on Senior/Staff engineering work.

Education

Moscow Institute of Radio Engineering, Electronics and Automation - Specialist Degree, Information Security of Telecommunication Systems, 2003-2008. Thesis: "Modern Website Attack Vectors Analysis."