Kateryna

Kateryna Brovkova
Python / Django Backend Developer
Batumi, Georgia - open to remote or relocation | [відкрити контакти](див. вище в блоці «контактна інформація»)
github.com/katerynabrovkova | [відкрити контакти](див. вище в блоці «контактна інформація»)
PROFILE
Python/Django backend developer who independently designed and shipped two systems end to end: a REST API with
queryset-level permission enforcement and an AI assistant grounded in real database records, and a Celery-based Telegram
bot with concurrency-safe deduplication and per-subscriber delivery tracking. Comfortable owning a feature from schema
design through testing (80%+ coverage) to CI-enforced type safety and linting, and Docker-based deployment. Looking for a
backend role where the Python/Django stack is a strong fit, remote or on-site, open to relocation.

TECHNICAL SKILLS
Languages: Python, SQL
Frameworks: Django, Django REST Framework, Celery, aiogram
Databases: PostgreSQL, Redis
API & Auth: REST, JWT, OpenAPI / Swagger
AI Integrations: OpenAI API, prompt engineering, grounded response generation
Testing: pytest, pytest-django, fixtures, parametrize, async testing (asyncio)
Code Quality: Ruff, mypy, django-stubs, type hints
Infrastructure: Docker, Docker Compose, GitHub Actions, Git, Render
PROJECTS
BookShelf API
github.com/katerynabrovkova/bookshelf-api | live API documentation
REST API for tracking a personal reading life: a shared book catalog, private shelves with reading progress, and reviews
unlocked only after a book is finished.
• Implemented JWT authentication with object-level permissions enforced at the queryset level; unauthorized access to
another user's shelf or review returns 404 rather than 403, so the resource's existence is never revealed.
• Built an AI assistant (OpenAI API) that answers catalog questions grounded in real database records: retrieves matching
books by keyword search with a random fallback when nothing matches, and returns a dedicated 503 response if the AI
service fails.
• Wrote a pytest suite of 19 tests covering permissions, business rules such as blocking reviews until a book is marked
finished, and parametrized edge-case validation, reaching 88% overall test coverage.
• Integrated Ruff and mypy into GitHub Actions CI, catching type mismatches before deployment.
Stack: Django REST Framework, PostgreSQL, JWT, pytest, Docker, GitHub Actions, OpenAI API, drf-spectacular

Job Vacancy Bot
github.com/katerynabrovkova/job-vacancy-bot
Telegram bot that monitors job boards and delivers new vacancies matching each subscriber's keywords within 5 minutes of
publication.
• Celery and Celery Beat poll two job sources on a schedule, with a 4-minute task expiration that prevents stale runs from
piling up if a worker falls behind.
• Tasks are chained so fetching always completes before delivery, keeping the digest consistent even when a run is
delayed.
• Deduplication enforced by a database constraint on source and external ID rather than application-level checks, keeping
repeated polling safe under concurrency.
• Delivery tracked per subscriber, so a delayed worker catches up instead of silently losing vacancies.
• Covered with 24 pytest tests, including async tests for delivery logic and parametrized tests for URL and title parsing
across edge cases, reaching 81% overall test coverage.
Stack: Django, Celery, Redis, aiogram, PostgreSQL, pytest, Docker Compose, GitHub Actions

LANGUAGES
Ukrainian: native | English: B1, actively improving