Tianhao Zhang
I find where the numbers are wrong, then fix the step that let them through.
Math graduate (B.S., University of Florida, cum laude), former math teacher and IT help desk assistant. I look for remote data quality, reporting and automation work. Below are six projects, each with the problem, what I did, and the numbers.
Work
Case study 1 · Public data analysis
Medicare Part D drug spending, 2020 to 2024: what drove the $90.0B increase
Public CMS data, 14,536 rows. I split the growth in gross Part D drug spending, drug by drug, into patients, prescribing volume, price per dosage unit and newly listed drugs, reconciled to the cent.
+$90.0B gross Part D drug spending, $198.7B (2020) to $288.7B (2024), +45.3%
Read the case study →Case study 2 · Reporting automation
Federal Contracts Weekly: a scheduled pipeline that turns USAspending.gov into a Monday briefing
A Python command-line tool pulls public federal contract data every week and builds a one-page HTML and Excel report: new activity, contracts coming up for rebid, trends and top recipients.
$242.9M net obligations across 243 contract actions in 12 weeks
Read the case study →Case study 3 · Data cleanup
Month-end cleanup: invoice PDFs and sales exports into one audited Excel workbook
A small Python pipeline turns a month of messy supplier invoice PDFs and two inconsistent sales exports into one clean Excel workbook, with every fix, drop and flag logged.
7,113 audit entries; 205 real problems found
Read the case study →Case study 4 · Data quality
A deal analysis tool, and the 214 prices nobody had checked
A local tool ranks resale offers by net margin. The top offer looked like +$107; the store pages showed prices far above what the tool used. I traced why and added a live price check.
214 offer-store rows priced from one listed number
Read the case study →Case study 5 · Automation
Running many AI work sessions from one inbox
One coordinating AI session assigns work to project sessions that each own their files. Every result comes back as a card in an append-only event log.
388 events in the first 28 hours
Read the case study →Case study 6 · Process
A rule-based pipeline for about 200 course documents
As a math teacher I built two full courses alone. A rules file, a required check-and-compile step and a log of corrections kept about 200 documents consistent.
2 full courses built alone (Algebra 1 and 2)
Read the case study →Planned · SQL
Olist e-commerce: SQL analysis
A SQL analysis of the public Olist Brazilian e-commerce dataset (orders, items, payments, reviews, sellers). Planned next project; write-up after it is finished.
PlannedHow I work
- Show what was checked. In the deal tool every price says when it was last checked, and unchecked rows say so.
- Fix the step that let it through. The wrong prices came from one step that copied a single number to every store, so the check went into that step.
- Write the rules down. Rules files, required tests and logs of corrections, to keep fixed mistakes from coming back.
- Use AI as a tool I control. I use Claude Code daily. I write the rules, the checks and what counts as done; it writes the code and the tests; a second model reviews every change; I check the results.
SQL (intermediate, MySQL) · Python · Excel (VLOOKUP/XLOOKUP, charts) · Google Sheets · MATLAB · Zendesk · LaTeX · Claude Code
Contact
LinkedIn · Resume on request.