Data extraction · Analysis · Technical documentation
I turn messy sources — PDFs, scanned pages, screenshots, spreadsheets — into clean data and documents people can actually use. Most of the material I work with was never meant to become data. Getting it out is the job.
Structured records pulled out of PDFs, scanned pages, screenshots and photos into clean Excel or CSV. Read by hand, not by a raw OCR pass.
Sales trends, customer segmentation and RFM, product concentration, returns and exception review — written up as findings, not just charts.
Manuals, operating procedures, SOPs and reports built from source material, with numbered steps, tables and annotated figures.
A scoring table that existed only as a set of screenshots was rebuilt into a structured dataset, then reconciled against a 238-page handbook and two hours of screen recording.
A full year of online retail transactions cleaned, segmented and analysed — with the conclusion placed before the detail.
A 238-page equipment handbook, a folder of interface screenshots and two hours of recording were turned into a document a new operator can follow unaided.
A free Blender add-on for baked cinematic camera moves and first-frame export for AI
video generation. The full version adds depth and normal passes and batch export.
View on GitHub →
Monitored core e-commerce metrics daily and diagnosed abnormal fluctuations. Rebuilt the performance reporting logic, cutting maintenance effort by 30%. Prepared and cleaned datasets for AI model iteration.
Coursework in Python, probability and statistics, data analysis and AI. Study was interrupted by two years of military service.
Daily training and duty, plus network and information security study. Awarded Outstanding Conscript and “Four Haves” Outstanding Soldier; part of a team recognised for data analysis at a cyber contest.