Self-taught developer since the pandemic. I build libraries that are as fast as I can make them, improving as I learn.

Everything is Python at the surface, but I write CUDA or C underneath when Python alone isn't fast enough.

Currently building a B2B personalized products company. I keep the business-specific code private, but share the general-purpose libraries I create along the way.

GPU-accelerated image compositing for Python.

• Zero Dependencies: Works with standard NVIDIA drivers (No CUDA Toolkit/Visual Studio required).
• Studio Quality: 32-bit float precision throughout the pipeline.
• Massive Speedups: 10-200x faster than CPU for blur, blends, and transforms.
• Smart Memory: Buffer reuse for zero-allocation video/batch processing.
• Rich Features - +40 operations (gradients, blend modes, effects...)

Full Documentation

Transform any Python function into a complete web app automatically. Just type hints and run().

• 454 tests passing - battle-tested reliability
• Zero configuration - working web app in 30 seconds
• Auto-generated forms, file uploads, dark mode, validation, plots, downloads...

from func_to_web import run

def divide(a: int, b: int):
    return a / b

run(divide)  # → http://127.0.0.1:8000

Full Documentation • PyPI Package

Turn Python functions into desktop GUI applications. The original version that inspired FuncToWeb.

• Cross-platform (Windows, macOS, Linux)
• Real-time updates or manual execution mode
• Built on Kivy

from functogui import App

def is_even(number: int = 4) -> bool:
    return number % 2 == 0

App(is_even)

Simple Python server deployment toolkit. Deploy to remote servers with just a few lines of code.

• Deploy FastAPI/Flask apps to $5 VPS servers
• Built on Fabric - mix with pure Fabric commands
• Supervisor service management

Fast G-code bounding box generator and framing tool in pure C.

• 26x faster than Python
• Zero runtime dependencies
• Handles 500MB+ files in seconds
• Universal compatibility (CNC, laser, 3D printer)

Fast Python logger with optional Telegram notifications. Separate channels for routine logs and critical errors.

• 195k logs/sec (Linux) / 80k logs/sec (Windows)
• Thread-safe with crash protection

Simple library for controlling CNC machines with GRBL firmware.

• Intelligent buffer management
• Auto-recovery from alarms
• Auto-disconnect detection

PyAlias Command aliases for Windows that actually work like native commands.

• 21x faster than .bat files
• Works in any terminal
• Zero startup overhead

Get selected files and folders in Windows. One line of code.

• Safe by design - read-only Windows Shell COM APIs, no keyboard simulation
• Works everywhere - File Explorer and Desktop
• Zero side effects - no clipboard modifications or system state changes

Pull requests welcome. I appreciate clear issues, minimal examples, and focused PRs.