For Staffly CRM, I built a CASL-compliant SMS outreach system running on AWS EC2. I also fixed matching quality in Staffly.ai's SANI recruiter-matching assistant.
I co-built CoKeeper, an ML transaction categorization tool with a Postgres double-entry ledger. It reached 92.5% accuracy on client data, with zero wrong auto-posts.
On my FX and equity research platform, I used leakage-safe, purged walk-forward validation on 21.7 years of OANDA data. None of the strategies passed the tests, so I reported the null results rather than shipping a bot.
I built Scraper 5000, a FastAPI and Playwright lead-generation console with 103 passing tests, and licensed it to its first reseller customer. At Shethynks Corporation LLC, I work in data analysis and technical operations.
