Multi-Broker Data Platform (2026)
Why it exists: Production-style backend & data infrastructure platform delivering automated ingestion, dbt modeling, and low-latency API metrics endpoints from scratch.
How it works: Ingests multi-source financial and telemetry APIs via containerized Apache Airflow 3 with asset-based cascade triggers. dbt Core models (via Astronomer Cosmos) transform raw datasets in PostgreSQL using incremental materialization and automated quality tests. Exposes analytics via a high-performance FastAPI & SQLAlchemy Core read layer without ORM overhead.
Tech used: Python, FastAPI, PostgreSQL, dbt Core, Apache Airflow 3, SQLAlchemy Core, Docker, REST APIs.
View Code on GitHub →
E-Commerce & Asynchronous Logistics Engine (2024–2026)
Why it exists: Connect internal retail systems with external delivery APIs (Yandex Eats/Delivery) for a nationwide pharmacy chain, creating a digital sales channel driving ~10% of total revenue.
How it works: Architected asynchronous processing pipelines using RabbitMQ and Python workers to synchronize orders and inventory status in real-time. Designed PostgreSQL integration views, custom validation schemas, and resilient retry handlers with dead-letter queues (DLQ) to ensure 99.9% fault tolerance against external API network timeouts and duplicate webhooks.
Tech used: Python, PostgreSQL, RabbitMQ, REST APIs, Django REST Framework, Celery, Docker.
Codebase is private under NDA - architecture and business outcomes are accurate.
Retail Data Infrastructure & Analytics Platform (2024–2026)
Why it exists: Provide a single high-throughput source of truth for financial reconciliation, stock control, and operational metrics across 2,000+ retail stores.
How it works: Designed and maintained a PostgreSQL Data Warehouse (400M+ rows in core fact tables) using table partitioning, indexed schemas, and optimized SQL logic. Automated daily ELT pipelines with dbt Core, implementing strict data quality contracts and testing to ensure DWH totals strictly match live transactional data.
Tech used: Python, PostgreSQL, dbt Core, SQL, Docker, Data Modeling.
Codebase is private under NDA - architecture and business outcomes are accurate.
