I take messy problems and
ship working software.

UK charity, ~£3.3M/year. When I joined, the technology was in fragments. Every donate button routed through an agency's hosted checkout the charity didn't own. Donations were scattered across three platforms with different payout cycles (2 days vs weekly vs two months). CRM integration was broken. Gift Aid reconciliation was manual. Mobile donation flow took over 30 seconds.

Over 18 months as the sole developer, I replaced every external dependency — 14 repositories. New checkout with Stripe PaymentIntents + SetupIntents, Apple/Google Pay, progressive disclosure, Zod validation. P2P fundraising engine with team pages, leaderboards, and URL attribution across 15 Prisma models. Config-driven campaign pages with a 30-field interface so the fundraising team could launch Ramadan appeals without a developer.

Then the data layer. Three sync pipelines: Enthuse via Playwright with a four-layer auth chain (SSO → magic-link email → Gmail OAuth → TOTP), LaunchGood via Playwright CSV export, and a MySQL→PostgreSQL warehouse with five data streams landing in raw tables then transforming into a canonical model. That model feeds N3O CRM queue payloads and a donor messaging system on n8n + Chatwoot + WAHA. Terminated the agency retainer (£1,200/month saved). Migrated hosting.

The charity needed to reach partner organisations across the UK — 100K+ registered charities in the Charity Commission dataset. The fundraising team was manually searching, manually qualifying, manually writing emails. No tooling, no scale.

I built a pipeline: raw Charity Commission data into PostgreSQL, then OpenAI translates natural language queries ("education charities with income over £1M operating nationally") into SQL filter logic via a custom segment engine. Qualified leads get personalised outreach — emails and talking points generated against each charity's actual profile, income band, and operational focus. Not mail-merge. Apify enriches contacts to find the CEO or Head of Fundraising. The whole thing runs from a CLI with deterministic Python underneath.

Charities using third-party pledge platforms were losing money to fees, waiting weeks for payouts, and had no control over the donor experience. The existing options — Enthuse, LaunchGood — were slow to pay out, inconsistent with Gift Aid handling, and charged admin fees the charity couldn't avoid. I built an alternative from scratch.

Full pledge checkout with Stripe integration — commit now, pay later, with three payment rails (bank transfer, GoCardless direct debit, card). Pledge lifecycle tracks through five statuses: new → initiated → paid → overdue → cancelled. Built-in bank statement reconciliation with configurable column mapping. But the real story is the AI "Steward": an embedded assistant that helps fundraisers manage their campaigns. It uses OpenAI with Gemini fallback, structured function calling with 13 tools that return rich visual cards — stats grids, pledge tables, campaign detail, WhatsApp status, nudge drafts. Every response is grounded in real pledge data.

I stress tested it across 31 scenarios in 10 categories — both bash and Python harnesses. Dashboard overview, pledge lookups, campaign queries, setup status, action requests, navigation, edge cases (gibberish input, dangerous requests like "delete all our data"), brand compliance (regex checks for forbidden terms: "AI assistant", "language model", "ChatGPT", "artificial intelligence"), persona scenarios (event lead, treasurer asking about Gift Aid, attribution queries, pledge amendments, trustee report prep), and tone validation (gives specific advice, responds with empathy not shame). 31 of 31 pass.