leonid
layout:
[RU]

Software Engineer

Leonid Matyushin

Moscow, Russia

inspect experiencedownload CV
Leonid Matyushin

career.md

experience

Nebius

Software Engineer

Aug 2023 – Nov 2025Amsterdam, Netherlands
  • HWaaS: Disk Replacement AutomationDesigned and implemented infrastructure that fully automated failed-disk replacement, in particular the creation of Jira tickets for data-center engineers. I built the solution as part of a Kubernetes operator and added a mechanism that disables automatic replacement when monitoring behaves anomalously. The system has performed 1,000+ disk replacements across 4+ regions for YDB and an in-house S3-compatible object storage. Extended the operator gRPC API, performed CRD migrations, agreed failure scopes and regional rollout order with SREs and managers, and set up monitoring and alerting.
    • Go
    • Kubebuilder
  • HWaaS: Automated DeploymentDesigned and implemented an automated Kubernetes build and deployment process for the HWaaS service, which ships as a set of tightly coupled Kubernetes resources. Adopted GitOps delivery so releases became reversible, faster, and transparent to other developers.
    • Kubernetes
    • TeamCity
    • Timoni
    • Helm
    • Argo CD
    • Argo Workflows
  • Dubformer: Media Processing BackendDesigned and implemented machine translation and text-to-speech microservices and integrated them into the automatic video translation pipeline. Developed REST APIs for individual services and for the pipeline as a whole.
    • Python
    • Airflow
    • FastAPI
    • MongoDB
    • AWS SQS
    • AWS S3
  • Dubformer: Infrastructure EngineeringBuilt monitoring and alerting that exposed backend behavior and load through dashboards and enabled rapid incident response through messenger notifications. Implemented CI/CD pipelines and reduced the total build and test time from 60 to 10 minutes.
    • Docker
    • GitHub Actions
    • Grafana
    • Prometheus

NtechLab

Software Engineer

Nov 2020 – Aug 2023Moscow, Russia
  • Speed and Memory Benchmarking ServiceDesigned and implemented an internal service for benchmarking inference speed and memory consumption of computer vision models developed in the R&D department. Exposed a REST API and built CLI and SDK interfaces on top of it. Proactively collected feedback and drove adoption across multiple teams, including object detection, face recognition, and license plate recognition, significantly reducing research iteration time.
    • Python
    • Celery
    • PostgreSQL
    • RabbitMQ
  • Data Labeling PipelinesDeveloped Airflow pipelines that automated data-labeling workflows for image-classification tasks in the R&D department.
    • Python
    • Airflow
    • Yandex Toloka
  • NSFW Image ClassificationDeveloped an NSFW content-detection model that outperformed popular open-source alternatives on public and internally created benchmarks, then deployed it to customer infrastructure.
    • Python
    • PyTorch
    • PyTorch Lightning
    • TorchServe
    • Albumentations
  • Falling People DetectionDeveloped a falling-people detection model using a domain-adapted neural network on top of an open-source skeletal-pose model. Achieved a high true-positive rate at an extremely low false-positive rate on benchmarks I collected myself, and owned the lifecycle from data collection and labeling through product integration.
    • Python
    • PyTorch
    • TensorRT
    • OpenVINO
  • Acceleration of the Facial Recognition ModelAccelerated the GPU face-recognition model by 10% without accuracy loss by using neural-architecture-search techniques to identify a better backbone.
    • Python
    • PyTorch

Laboratory of Methods for Big Data Analysis

Research Intern

Jan 2020 – Nov 2020Moscow, Russia
  • Conducted seminars for a third-year undergraduate Data Analysis course and received a 4.5/5 student rating.
  • Assisted with seminars at the Sixth Machine Learning in High Energy Physics Summer School.
  • Built an oil-production prediction model for the Saudi Aramco Moscow office.

Deeplight Ventures

Software Engineer

Feb 2019 – Feb 2020Moscow, Russia
  • Won second prize in an oil-reservoir image-segmentation competition.
  • Built a pump-failure prediction model.
  • Implemented an optimization procedure for well interventions on an oil field.

Sberbank

Data Science Intern

Aug 2017 – Sep 2017Moscow, Russia
  • Designed and developed a dataset for an NLP contest.