Yaroslav

Contacts
[відкрити контакти](див. вище в блоці «контактна інформація»)
Yaroslav Bordovyi
[відкрити контакти](див. вище в блоці «контактна інформація») LinkedIn
Python Developer
Telegram GitHub

Summary

Software developer with two years of commercial experience building backend services in Python. I work on a production
data-annotation platform used by distributed annotation teams, and take features end to end — architecture discussion,
implementation, code review, rollout, and profiling under real load. I co-developed a microservice toolset on FastAPI with
Celery, Kafka, Redis and PostgreSQL, and work daily with MongoDB aggregation pipelines and schema migrations on the
main product. Beyond pure backend, I ship what the product needs — TypeScript/React on the client, FFmpeg-based media
pipelines, integrations with external annotation systems — and one of my streaming optimizations cut editor memory usage
roughly fivefold on low-end hardware. I value code that survives review and solutions verified against real data rather than
assumptions.

Experience
Software Developer · Keymakr / Keylabs · March 2025 – present
Platform product development

Video streaming for the annotation editor (in-browser): implemented WebCodecs hardware decoding with
mediabunny demuxing and an ffmpeg.wasm software fallback, GOP caching via the Cache API, server-side frame
maps, and HTTP Range media serving - removing the pre-processing step that used to split every video into individual
frames.
Profiling harness for the streaming pipeline: used it to drive optimization - 480p decode during playback with full-
resolution frames on pause - cutting editor memory usage roughly 5x and giving stable 1x-4x playback on low-end
hardware.
Server-side transcoding tier: authored the architecture - FFmpeg H.264 mezzanine, HLS / fMP4 segments, hot/cold
caching with TTL policies, and a MongoDB job queue with a worker daemon - backed by encoding benchmarks,
approved by the team; implementation in progress.
Operator worklog subsystem: reworked multi-tab shift and work-time synchronization, closing beacons, and correct
billing across shift boundaries, plus MongoDB schema migrations; chose a ping-based protocol over a WebSocket
prototype after it proved more reliable under tab suspension and connection loss.

Backend services & integrations

Internal microservice toolset: developed a project cloning service on FastAPI with Celery workers, Kafka messaging,
Redis, PostgreSQL with SQLAlchemy and Alembic migrations, JWT authentication, and job-history dashboards -
deployed with Docker + nginx + supervisord and monitored in New Relic.
Cross-platform annotation migration: built bidirectional transfer of media and annotations between the Keylabs
platform and CVAT, preserving object identity, attributes, and frame timelines.
Pre-annotation models: fine-tuned pre-trained detection models (YOLO family) on in-house datasets when out-of-the-
box quality proved insufficient, and integrated the improved models into the pre-annotation pipeline; integrated Voxel51
(FiftyOne) and deployed open-source annotation tools for isolated client workflows.

Data pipelines for enterprise clients

Annotation transfer across decimated frame sets: resampled interpolated timelines instead of copying keyframes,
handling non-uniform decimation between projects - verified frame by frame, zero object loss.
Orthophoto annotation stitching: reassembled sliced annotations onto ~16 000 x 19 400 px images and converted
them to client-format GeoJSON (EPSG:6991) using a pure-Python GeoTIFF tag parser.

Technologies: Python, FastAPI, Celery, Kafka, Redis, PostgreSQL, MongoDB, SQLAlchemy/Alembic, TypeScript, React,
PHP, FFmpeg, Docker, nginx, AWS S3, GCS, New Relic, Bitbucket Pipelines.
Python Developer · September 2024 – February 2025

Designed and developed RESTful APIs using Django REST Framework and FastAPI.
Integrated Docker for containerization and deployment of applications.
Automated data extraction and routine tasks with Scrapy and Selenium.

Technologies: Python, Django REST Framework, FastAPI, PostgreSQL, Redis, Docker, HTML/CSS, Scrapy, Selenium.

Skills
Python: Python 3.11+, advanced Python (decorators, iterators, generators), asynchronous Python (asyncio,
concurrency, parallelism)
Backend & APIs: FastAPI, Django, Django ORM, Django REST Framework, SQLAlchemy, Alembic, REST API
design, JWT authentication
Async processing & messaging: Celery, Kafka, Redis, job queues, long-running worker daemons
Databases: PostgreSQL, MongoDB (aggregation pipelines, indexes, schema migrations), SQLite, Redis
Infrastructure & DevOps: Docker, Docker Compose, nginx, supervisord, uWSGI, CI/CD pipelines (Bitbucket
Pipelines, GitHub Actions), AWS S3, Google Cloud Storage, New Relic, Linux / bash
Performance engineering: profiling, benchmarking, memory and latency optimization for low-end target hardware
Media processing: FFmpeg, WebCodecs, HLS / fMP4, H.264 / VP8 / VP9, frame-accurate video playback
ML & CV tooling: model fine-tuning (YOLO family), Segment Anything (SAM 2.1 / 3) integration, Roboflow, Voxel51
(FiftyOne), Label Studio / CVAT deployment and integration
AI-assisted development: Claude Code, ChatGPT, Gemini, MCP servers, agent-based code review and workflow
automation
Web technologies: TypeScript, React, HTML, CSS
Other languages: PHP 8.2 (reading and maintaining a custom in-house framework)
Data formats: COCO, YOLO, GeoJSON, GeoTIFF, coordinate reference systems (EPSG)
OOP & algorithms: SOLID principles, design patterns, complexity analysis
Data analysis & scraping: Pandas, Matplotlib, BeautifulSoup, Scrapy, Selenium
Collaboration & process: Git, Bitbucket, GitHub, code review, Jira, Agile (Scrum, Kanban)
English: Intermediate

Education
Oles Honchar Dnipro National University
Master's degree, Ecology · 2024 – 2026
Bachelor's degree, Ecology · 2020 – 2024 — diploma with honors

Dnipro Basic Professional Medical College
Specialty "Obstetrics" · 2015 – 2018 — diploma with honors

Additional education
Python course at Mate academy (OOP, Python, Django, FastAPI)
Python course at Hillel IT School (certificates 75905064, 61846679)

Achievements
Candidate Master of Sports in Figure Skating
Published a scientific paper during university studies