Godfrey Osagiede
Software Engineer — Payments & AI Systems
Dallas–Fort Worth, TX · open to NYC · remote-friendly · (469) 660-7278 · godfreystorm@gmail.com · linkedin.com/in/godfrey-osagiede-a29013144 · github.com/godfreystorm · FocusDown on the App Store · throwda.com · curbtip.com · godfreystorm.github.io/lab
Payments-focused software engineer who owns Adyen operations across 8,000+ stores in production and ships products end-to-end outside of work — an iOS app with 10,000+ users, an NFC tap-to-tip payments product, and a production-grade WhatsApp fintech. Also run agent systems as measured experiments: a public evolution lab that ran 41 LLM-agent colonies over 10,810 generations and 356 pre-registered A/B trials, and published the result that disproved its own founding thesis. Deep production database experience (MongoDB, MySQL, Postgres).
Payments & Data Operations Engineer
LineLeader
Childcare-management SaaS serving 2,500+ franchise organizations across US, AU, and UK.
- Primary engineer for all Adyen payment operations (Balance Platform, Management, Legal Entity APIs): store configuration, fee structures, KYC enablement/resets, and sweep management across 8,000+ stores.
- Built a Node.js audit-and-repair tool that validated 8,651 Adyen stores across two production environments, cut indeterminate cases from 2,393 to 87 via Legal Entity API lookups, and safely repaired 4,088 misconfigured stores under a compliance deadline, with rate-limit retry/backoff.
- Leading a timezone migration touching 30–40M records across 2,905 organizations (in code review).
- Built the superadmin billing epic that lets staff set Adyen platform fees in-app, retiring an entire class of engineering ticket — fee changes were previously out-of-band DB edits. Shipped the tabbed shell, the fee table, and a permission-gated
updatePlatformFeesmethod with server-side validation; now extending it to onboard organizations onto Adyen sub-accounts from the UI, replacing scripts run by hand today. - Wrote ~570 lines of unit tests against ~1,340 lines of feature code on that epic, shipping design doc → plan → code → tests → PR review.
- Operate daily in production MongoDB (aggregation pipelines,
$graphLookupover 2,500+ org hierarchies, bulkWrite backfills) and MySQL (284-table CRM schema, FK-cascade-safe bulk operations, 600K+ row inserts) across multiple environments. - Every production mutation follows dry-run → verified backup → approval → apply → count verification, shipped through Bitbucket PR review.
Software Engineer (Products)
Raidersoul LLC
Build and operate three products end-to-end alongside my full-time role — one earning on the App Store, two payments products built and deployed, pre-launch.
- Grew to 10,000+ users and ~1,500 monthly actives in 5 months, some on paid subscriptions; 510 commits solo.
- Wrote ~8K lines of native Swift: WidgetKit home-screen widgets, Live Activities/Dynamic Island, and a Screen Time (Family Controls) app-blocking stack.
- Built the full monetization funnel — StoreKit + RevenueCat + remotely configurable Superwall paywalls with webhook processing on Supabase edge functions — and instrumented a 20+ step onboarding funnel in Mixpanel, fixing the #1 drop-off (22.6%) found in the data.
- Automated growth operations with AI agent pipelines: daily content generation/posting, creator outreach, and a contracted influencer program (10 signed creators).
- Riders tap an NFC tag and tip in one Apple Pay confirm — no app, no signup; the driver keeps 100% and revenue comes from a rider-side service fee.
- Built the full money path end-to-end: phone-OTP driver auth, tag/kit activation lifecycle, server-priced Stripe PaymentIntents, and an idempotent webhook ledger verified against replayed and deliberately dropped events.
- Set the fee structure from a choice model calibrated on a published 12M-trip pricing experiment, after re-validating the category's tipping assumptions against 20M NYC rideshare trip records — the modeling overturned four of the five stats the original spec was built on.
- Built a SwiftUI App Clip (NFC lock-screen invocation → Apple Pay, zero install) and a React Native/Expo driver dashboard; wrote a Puppeteer audit harness covering 19 routes × 2 color schemes × 4 widths for WCAG contrast and tap-target compliance.
- Bill payments, P2P, and community collections entirely inside WhatsApp; launch-partner dashboard seeded with a Lagos church's real 84-account chart of accounts.
- Implemented double-entry bookkeeping with DB-enforced balanced journals, bank reconciliation from real bank CSV/XLSX statements, and financial reports (balance sheet, cashflow, budget variance) — ~45K LOC dashboard with 75+ test files (Vitest + Playwright).
- Designed a direct-charge payment architecture (Paystack
charge_authorization, zero stored value) that avoided a ₦100M PSSP licensing requirement; multi-provider routing (VTPass/eBills) with margin-based failover and SQL kill switches.
Ratchet Lab — LLM agent evaluation harness & open-endedness research
Independent research on whether populations of LLM-written agents accumulate capability across generations.
- Built the full evolution harness (Node.js, SQLite archives, multi-provider LLM routing) and ran 41 colonies, 10,810 generations, 30,287 API calls, and 356 A/B trials for ~$30 total compute.
- Pre-registered every run behind a Fisher significance gate after an earlier trial arm proved statistically empty; every run ships raw archives, spend ledgers, and full match replays.
- Published the ablation that disproved the thesis the project was built on — the no-library control beat the shared-library arm — plus two retractions of my own earlier claims.
Junior Software/IT Engineer
RHD Industries
- Built an AI route-optimization desktop tool (Python, Google Places API, OpenAI API) that sequenced multi-stop delivery routes, reducing per-shift drive time.