One file
Positioning
Not “AI magic.” Boring business work made repeatable.
I focus on bounded projects where a working deliverable can be tested quickly: parse files, clean data, create PDFs, sync records, generate reports, and package the result so a non-developer can run it again.
Fixed-price service · CSV + XLSX
Spreadsheet Rescue
Send the messy file. Get back a cleaned copy plus a plain-English quality report—without changing your original.
41/41tests passing
What gets fixed
- Confusing headers normalized
- Accidental surrounding spaces removed
- Exact duplicate rows removed
- Blank cells counted by column
- Clean CSV/XLSX + JSON quality report delivered
Up to three files
Repeatable workflow
Per-file limits: 10,000 data rows, 1,024 columns, and 1,000,000 cells. XLSX cleanup is text-only; formulas, formatting, macros, and extra worksheets are not preserved. Async-only—no calls or screen sharing.
Proof image uses fictional sample data and reserved example.com addresses. No client data or sale is claimed.
Selected work
Verified project examples
These are local proof projects built with fake/sample data only. They show the kind of client work I can adapt after project kickoff.
Python · Excel · parsing · tests
Email-to-Excel Sales Data Automation
Reads saved email bodies, extracts shop/location, date, sales amount, and customer count, then writes clean rows into an Excel workbook with JSONL audit logging.
- Verified demo: 3 passing tests
- Output: email_sales_log.xlsx
- Production path: Outlook/Gmail export, IMAP, Make/Zapier, scheduled Windows run
Airtable · PDFs · invoices
Airtable-style Invoice PDF Generator
Converts structured order and line-item data into a professional invoice PDF. Designed to map cleanly to Airtable API records after client access.
- Verified demo: 3 passing tests
- Output: sample_invoice.pdf
- No private data or credentials
Excel · CSV · safe dry-run
Excel-to-WhatsApp Message Automation
Reads contact spreadsheets, validates required columns, personalizes messages, normalizes phone numbers, and generates safe dry-run review output before any live send.
- Verified demo: 4 passing tests
- Output: dry_run_messages.csv
- API-ready without exposing secrets
Browser game · UX · controller-first prototype
Odd Lot Pawn Prototype
A playable browser prototype for a pawn-shop deal-making roguelite. Shows product thinking, UI polish, gamepad support, local hosting, and iteration from idea to testable artifact.
- 7-day run loop, cash/rent/reputation systems
- Gamepad + keyboard controls
- Mobile/LAN playtest flow
Godot · Windows build · game systems · QA automation
Zero-Day Defender: Firewall Breach
A Godot cyber-survivor prototype with auto-fire combat, malware waves, XP/card upgrades, persistent Lab upgrades, Bits economy, settings, credits, jukebox, and export verification.
- Playable Windows export: source + exported autotests pass
- Implemented loop: movement, targeting, upgrade cards, boss spawn, meta progression
- Release planning: Steam readiness audit, playtest scorecard, packaging roadmap
Services
Good-fit project types
01
Spreadsheet automation
Excel/CSV cleanup, formulas, generated workbooks, recurring reports, and audit-friendly outputs.
02
File and email parsing
Turn recurring emails, text files, JSON, or exports into structured data that can be reviewed and reused.
03
API glue scripts
Small connectors for Airtable, Google Sheets, CRMs, invoice tools, and other business systems.
04
Proof-first prototypes
Fast demos with fake data, tests, screenshots, and a clear path to production after scope is confirmed.
Working style
Async, scoped, and testable.
- Define the output. What file/report/API result should exist when the job is done?
- Build a small working version. Use sample data first, then adapt to real formats after access is granted.
- Verify it. Run tests, produce demo output, and document how to repeat it.
- Hand off cleanly. Short written instructions, config templates, and safe handling of credentials.
Contact
Need a small automation built?
Best fit: async projects with clear inputs and outputs. I prefer written requirements, sample files, and small milestones over live calls or screen-share sessions.