AI / LLM Engineer (Claude): Data Cleaning and Matching Pipeline

CharityHowTo is an online training and EdTech platform for nonprofits, running for 18 years across four brands: CharityHowTo, Nonprofit Library, Nonprofit Apps and Nonprofit Discounts. We created a service called Donor Data Clinic (https://donordataclinic.net) and ran a successful beta with real nonprofits. I'm looking for one independent AI engineer to build and run the technical side of it. This is an individual role, not an agency. ⠀ You own delivery. CharityHowTo brings the customers through the marketing we already do across those brands. ⠀ What the service does: a nonprofit sends a donor file exported from their CRM (Bloomerang, DonorPerfect, Little Green Light, Salesforce NPSP, Neon). They get back a clean file, with standardized addresses, validated emails and phones, and duplicates, households and deceased donors flagged. They also get a fundraising plan: lapsed donors, top donors, and matches between their donors and foundation boards using IRS 990 officer and trustee lists. Turnaround is 72 hours to start, and the goal is close to instant. ⠀ Where it stands: it's not live yet because the delivery side needs to be rebuilt, and that's the job. ⠀ What you'd do Build the pipeline that turns a messy CRM export into the clean file and the fundraising plan Decide which steps go to an LLM and which stay plain code Put a person's review where a match is uncertain Keep it running and improve it from customer feedback ⠀ Most of the work is the build up front, then steady improvement after that. This is ongoing work, for as long as it suits both of us. ⠀ Who I'm looking for You build working AI systems, not just demos. You know how to get reliable results from an LLM Hands-on experience with Claude is required (the API, structured output, or Claude Code) Strong Python, and data pipelines you've run in production Real experience matching or cleaning messy records You handle private data carefully. Donor files hold personal information Clear written communication. We work mostly remote and async Comfortable working alone Nonprofit or donor-data experience is a plus, not a requirement ⠀ Commercial structure is flexible: revenue share, paid contract, or a hybrid. We'll talk it through on a call and agree it in writing before any work starts. ⠀