Особисті дані приховані

VITALIY LEHKII
PYTHON BACKEND DEVELOPER
[відкрити контакти](див. вище в блоці «контактна інформація») GitHub LinkedIn

ABOUT ME
A proactive and detail-oriented Python back-end developer with a strong commitment to writing clean, maintainable, and
efficient code. I approach challenges methodically, focusing on understanding the core problem to deliver robust and scalable
solutions, often leveraging containerization and microservice concepts. I am a collaborative team player who believes that
sharing knowledge and supporting colleagues is key to a project's success. Eager to contribute my skills to building high-
quality products andgrow withina professional team.

EDUCATION LANGUAGE
TARASSHEVCHENKO NATIONAL UKRAINIAN -NATIVE
2023 - now UNIVERSITY OF KYIV ENGLISH - INTERMEDIATE+
SOFTWARE ENGINEERING

WORK EXPERIENCE
PYTHON BACKEND ENGINEER | EON+ AI | 3 MONTHS
Led the migration of legacy microservices from PHP/TypeScript to Python (FastAPI), enabling ~10x higher concurrent I/O and
reducing p95 latency by ~30% for IO-bound workflows.
Designed a resilient event processing engine using Redis Streams (consumer groups, DLQ) that sustains ~400 events/sec per
pod with strict idempotency.
Optimized fault tolerance: implemented an XAUTOCLAIM strategy to reclaim stuck messages in ~33s (worst-case) — ~6x-18x
faster than standard restart-based recovery (3-10 min).
Engineered self-healing consumers with Exponential Backoff & Jitter and Graceful Shutdown to minimize message loss during
deployments.
Integrated LLM & Voice AI (OpenAI, ElevenLabs): built a WebSocket audio bridge with <20ms server overhead, achieving a
typical end-to-end voice latency of ~0.8s.
Improved observability by implementing structured logging and health probes to track retry rates and operational signals for
rapid incident debugging.

PET PROJECTS
TWILIGHT STORE , LINK (DJANGO, REDIS, POSTGRES, DOCKER, NGINX)
Implemented a server-side Redis caching layer that reduced catalog page load time by 60% (from 300ms to
120ms).
Optimized database performance to solve the N+1 query problem, significantly reducing SQL queries on the catalog
page.
Built an automatic cache invalidation mechanism using Django Signals to ensure 100% real-time data consistency.
Containerized the application by configuring Docker Compose to manage 4 isolated services (Django Gunicorn,
PostgreSQL, Nginx, Redis).
Configured Nginx as a reverse proxy and static file server, which reduced the load on the Django server by 35%.
Implemented server-side logic for dynamic, multi-parameter product filtering (e.g., by category, size).
Integrated Cloudinary CDN for media file storage, saving ~80% of server disk space.
VIDEO PROCESSING API, LINK (FASTAPI, CELERY, REDIS, DOCKER, GOOGLE CLOUD)
Developed a RESTful API using FastAPI to automate video creation and mixing workflow.
Reduced API latency from minutes to <200ms by implementing asynchronous background processing with Celery & Redis,
ensuring non-blocking user experience.
Integrated 3rd-party APIs: connected ElevenLabs for text-to-speech generation and Google Cloud Storage for storing result
files.
Built a scalable file storage system that supports switching between local and cloud storage via environment variables.
Optimized worker stability by isolating heavy tasks, which solved potential memory leaks during video processing.
Containerized the application (API, Worker, Redis) usingDocker Compose for easy deployment and development.

SKILLS
Python(Django, FastAPI) PostgreSQL Git, Github Actions
JavaScript Redis (Streams) RestAPI
Pydantic Celery JWT/OAuth2
Numpy Docker, docker-compose HTML, CSS
Object-Oriented Programming (OOP) Nginx SOLID, DRY, KISS
LLM (OpenAI, ElevenLabs) Google Cloud (GCS) WebSockets