Building scalable backend services, OCR/Document Intelligence pipelines and developer-first tooling
Repos • Tech stack • Projects
- 🐍 Backend engineer with a strong Python background and hands-on experience in ML-driven systems
- 📄 Focused on document intelligence: OCR/VLM pipelines, PDF parsing, layout detection, text extraction
- 🧱 I build systems that are stable, observable and deployable - not just correct in theory
- ♻️ Work across the stack: data preprocessing, model inference, API design, infra and CI/CD
- 🔬 I value benchmark transparency, reproducibility and engineering discipline
- 🧠 Interested in architecture, clean boundaries, maintainability and developer experience
- 🌐 Experienced with GPU-based workloads and high-throughput service pipelines
-
🧾 xcells
Utility library for working with Excel - clear API and practical functionality. -
🤖 Integram
A FastAPI-based REST API for working with metadata and dynamically constructing SQL queries for PostgreSQL. The system provides RESTful endpoints for retrieving information about terms, objects, attributes, and reference tables, as well as creating and executing complex queries with automatic JOIN generation. The application uses SQLAlchemy for asynchronous database operations, includes JWT authentication and route-protection middleware, and is deployed via Docker Compose. A key feature of the project is the dynamic QueryBuilder, which analyzes relationships between tables and automatically generates optimized SQL queries based on the requested columns. -
⚙️ python-module-compiler
Tool for compiling Python modules into.pycwhile preserving structure. -
🌐 python-sitemap-generator (fork)
Experiments with refactoring patterns and tooling improvements.
More experimental repos: benchmarks, infra tests, data utilities - feel free to explore 👀
Open to conversations about:
- production Python backends
- document intelligence / OCR pipelines
- GPU-powered inference setups
- architecture, performance and reliability
Telegram: @horhe_dvlp
Feel free to reach out or open an issue.
Thanks for stopping by 👋

