Dmytro Milinevskyi - Distributed Systems

Turning ambiguous production problems into
boringly reliable and efficient systems.

Distributed systems engineer focused on reliability and operability—from embedded systems and distributed storage to high-throughput, real-time data processing systems and control planes. I work across teams on system design, production delivery, and the processes and tooling needed to operate and evolve those systems safely.

Define the behavior, design for failure, validate the change, and leave the system easier to operate

I make the decision explicit: the production constraint, expected behavior, technical trade-offs, cross-team dependencies and ownership, and evidence that will validate the approach.

I carry the change through end-to-end tests, live-traffic validation where appropriate, staged rollout, tested recovery procedures, and operational handoff.

I turn recurring operational work that requires expert intervention into tooling and validated, automated self-service workflows so other engineers can operate and extend the system independently.

Where this approach helps most

I help teams make high-throughput production systems more reliable, diagnosable, and easier to operate—from processing pipelines, routing and control planes to distributed storage.

Selected Evidence

Routing control plane for safe, self-service production change

Designed a metrics routing control plane that replaced static, specialist-managed configuration with validated APIs, transactional state, staged rollout, recovery procedures, and reusable migration workflows, enabling safer, repeatable production migrations.

Cross-domain monitor correctness during partial degradation

Worked across Alerting and Metrics to make monitor evaluation aware of ingestion-to-query health, reducing the risk that delayed or incomplete platform data would be interpreted as customer behavior; the framework supported safer degraded-mode recovery and was extended to additional product domains.

High-throughput realtime processing with ~3× lower resource use

Owned and optimized routing and processing systems handling billions of events per second across petabyte-scale data and thousands of workers, reducing resource use by ~3× through caching, sharding, and hot-path optimization while validating correctness and failure behavior under production conditions.

Distributed storage and synchronization

Stabilized correctness-sensitive distributed filesystem and synchronization internals, and productized the software so users could diagnose and recover from filesystem and service failures.

Embedded platform reliability

Owned firmware, kernel, boot, and recovery work that made embedded wireless platforms safer to update, diagnose, and operate remotely.

Career History

Dec 2017 - Sep 2026

Datadog · Paris, France

Distributed Systems Software Engineer

Built and operated routing control planes and high-throughput processing systems for Datadog's Alerting and Metrics products. Worked with Product and platform teams on monitor behavior, telemetry diagnostics, and production migrations.

Enabled self-service production migrations by authoring the design and leading the evolution of the metrics routing control plane, with staged rollouts, live-traffic validation, automated health gates that replaced manual operator checks, and simple recovery procedures.
Defined customer-facing monitor semantics and operator-facing telemetry diagnostics with Product, engineering leadership, and platform teams. Preserved compatibility for established monitor workflows while clarifying recommended evaluation behavior. Redirected a high-cardinality telemetry investigation proposal from a costly standalone pipeline to a bounded investigation workflow, and led its design and production delivery.
Reduced false positives and false negatives and made recovery more predictable during partial degradation by integrating data-freshness and completeness signals into Metrics ingestion, processing, storage and query systems, then using those signals in the Alerting evaluation pipeline.
Reduced resource usage across two high-throughput realtime processing systems by applying system-specific optimizations—caching, sharding, concurrency tuning, and hot-path optimization: ~10× for service-check ingestion, processing, and evaluation handling millions of events per second, and ~3× for the Metrics routing layer processing billions of events per second across petabyte-scale data and thousands of workers.
Enabled engineers outside the control plane team to complete realtime storage-tier migrations without direct support by building reusable migration tooling, AI-assisted troubleshooting, and end-to-end validation.

Apr 2015 - Dec 2017

Lima · Paris, France

Core Software Engineer

Built core components of a distributed filesystem, synchronization subsystem and manufacturing tooling.

Improved replication robustness, transaction performance, and garbage-collection efficiency so the distributed filesystem could handle very large trees with hundreds of thousands of files and directories.
Designed the factory test and provisioning process, built the software for hardware testing, firmware flashing, and device identification, and installed the system on-site.
Designed and implemented file sharing via public web gateways.

Dec 2008 - Apr 2015

Sequans Communications · Paris, France

Embedded Software Engineer

Developed LTE firmware and embedded platform software, including Linux and real-time kernels, boot and recovery, device drivers, and build infrastructure.

Implemented JFFS2 for vxWorks in read/write mode to simplify transition from vxWorks to eCos kernel.
Optimized USB, networking, and inter-processor layers, bringing throughput closer to hardware limits.
Made boot, recovery, and firmware updates more resilient. Built tools for post-crash diagnosis and reproducible multi-repo firmware builds.

Let's work together

Dmytro Milinevskyi

Distributed Systems · Technical Leadership · Paris, France or Remote

Turning ambiguous production problems into
boringly reliable and efficient systems.