Systems built to be reused.

Selected engineering work across backend architecture, AI systems, machine learning, and quantitative development. The focus is not only on applications, but on building software foundations that can be extended, tested, packaged, and reused.

01 · Open Source Python Library v0.1.0 · Published

FastApiBackbone

A reusable backend foundation for Python services built around FastAPI. Rather than treating FastAPI as the entire backend architecture, FastApiBackbone supplies the surrounding infrastructure required to build durable applications: environment-driven configuration, asynchronous SQLAlchemy, database lifecycle management, Alembic migrations, structured logging, health probes, and a reusable application factory.

Now available as a Python package. Installable distribution · semantic versioning · automated build and release workflow

FastAPI → API composition → application services ↓ configuration + lifecycle + structured logging ↓ SQLAlchemy 2.x ↔ PostgreSQL / SQLite ↓ Alembic migrations

PythonFastAPISQLAlchemy 2.xPostgreSQLSQLiteAlembicPydanticstructlogAsync I/OPyPI

View source → View on PyPI →

02 · AI SystemsActive Development

PAssist

A provider-agnostic AI platform built around the same backend engineering principles. PAssist combines a modular FastAPI service with identity and JWT authentication, persistent conversations, AI provider abstraction, local Ollama integration, retrieval, tools, memory, and extensible domain boundaries.

Identity → JWT authentication → sessions ↓ FastAPI application → domain modules ↓ AI provider contract → Ollama / compatible APIs ↓ SQLAlchemy 2.x → Alembic → PostgreSQL

PythonFastAPIJWTArgon2SQLAlchemyPostgreSQLOllamaLLMsRAGTool Calling

Explore PAssist →

The engineering philosophy

Infrastructure first. Domain second.

FastApiBackbone is the reusable infrastructure substrate: asynchronous application structure, database lifecycle, migrations, configuration, logging, health checks, and clean extension points. It is deliberately domain-neutral so that another Python service can adopt the foundation without inheriting assumptions from a particular product.

PAssist demonstrates the next layer: a concrete application domain built with identity, security, AI provider abstraction, conversations, tools, knowledge retrieval, and memory. Together, the projects show the progression from reusable backend infrastructure to a complex AI-enabled application.