Danyal Zia Khan
Python developer building automation, 3D reconstruction, and AI systems that replace manual work
Most of what I build exists because nothing off the shelf did the job. Pipelines that turn blueprints and reference images into scale accurate 3D environments. Scrapers that pull data from sites built to stop exactly that. AI models running on a client's own hardware instead of someone else's API. Six plus years of Python, and more than 200 scrapers delivered.
What I do
- 3D reconstruction and parametric modeling. Blender addons and Python tools that turn blueprint measurements and reference images into scale accurate environment structures such as towers, bridges, and roads. Producing a variant takes a change of input values instead of a modeling session, and AI assisted passes check the geometry for faults before it reaches an artist.
- AI assisted automation. Systems built on open weight models and run locally, so client data stays on client hardware: image understanding, object detection and dimension estimation, OCR across scans and technical drawings, and reasoning through llama.cpp.
- Web scraping. A Playwright based framework covering authentication, anti bot handling, proxy rotation, and parallel collection, developed across more than 200 delivered scrapers.
- Process automation. Python programs that absorb the repetitive work: large Excel and tabular processing, grouping and aggregation, data cleanup, bulk file handling, and scheduled jobs that run without anyone remembering to start them.
Recent work
Digital Twin Mesh Processing Pipeline
A production pipeline that turns generated 3D models into geometry artists can actually work with.
Parametric Structure Generation
Blender tooling that builds towers, bridges, and similar structures from blueprint measurements instead of modeling each variant by hand.
Product Similarity Search with Vision Models
An image based product matching system that compares catalog pairs at a scale nobody could review by hand.
Review Scraping and Sales Ratio Analysis
A scraping and analysis pipeline that estimates relative sales performance from public review data.