GET /giorgi 200 OK
about

I build backend systems that stay correct under load: workflow state machines, idempotent async pipelines, webhook delivery, PostgreSQL performance. 3+ years across marketplaces and identity verification.

Beyond core backend, I integrate LLMs with the same engineering discipline — an LLM call is an external dependency that needs retries, caching, monitoring and output validation. Not a magic box.

stack
core
Python 3.11FastAPIDjango / DRFSQLAlchemyPydantic
data & messaging
PostgreSQLRedisKafkaRabbitMQCeleryClickHouse
infra & quality
DockerKubernetesCI/CDpytestmypy
ai / llm
Claude APItool use / agentsMCPSSE streaming
projects
PDFer
AI-powered PDF data extraction — FastAPI + Claude API with retries, Redis caching, Prometheus metrics and a hand-rolled agent loop.
IoT Fleet Tracker API
Async telematics backend — FastAPI, SQLAlchemy 2.0 async, asyncpg; fuel-theft detection over vehicle telemetry.