Some implementation details are omitted under NDA. The information presented here comes from publicly available World Bank and Spatial Collective sources.
Digital Public Works Platform (KISIP II)
Spatial Collective / World Bank KISIP II · 2024 to Present
I manage the enterprise platform that coordinates digital data collection, automated stipend payments, and real-time reporting for a World Bank-funded youth employment programme. The system handles 50,000+ daily API requests across a 76-table PostgreSQL database.
The Situation
When I joined Spatial Collective, the World Bank KISIP II project was running on spreadsheets. 282 youth mappers across three Nairobi settlements — Embakasi Sokoni, Kahawa Soweto, and KCC — were submitting field work through a patchwork of Excel files and paid database subscriptions. Task tracking was manual, stipend calculations took days, and there was no way to verify data quality at scale. The World Bank needed real-time reporting, the government needed compliant data handling, and the field team needed something that just worked.
My Role: I was brought in to build and maintain the central platform that ties the entire operation together. That means syncing data from four external collection systems, running automated stipend calculations (base pay + quality bonus + overtime), enforcing 19 data integrity checks, and producing dashboards for World Bank and government stakeholders — all while keeping response times under 200ms for 200+ concurrent field workers.
What I Built
I built a full-stack TypeScript platform deployed on government-allocated server infrastructure with a 76-table normalized PostgreSQL database managed through Prisma ORM. The system pulls field data from ODK Central (mobile surveys), JOSM (building digitization), a microtasking bridge (AI consensus validation), and a learning platform — all unified through what I call a Work Ledger pattern. The payment engine calculates stipends using configurable formulas across different settlement types. I implemented an AI Guardian system that runs 19 integrity checks with health scoring and auto-remediation, sending email alerts through Resend when anomalies are detected. The visualization layer uses MapLibre GL with Supercluster for real-time geospatial dashboards. In 2025, I supported the expansion to four additional settlements: Kwa Rhoda and Karagita in Nakuru, Kaloleni and Obunga in Kisumu.
Data Pipeline
System Architecture
Geographic Coverage
Measured Impact
The platform now handles 50,000+ daily API requests with average response times under 200ms. It has processed KES 2M+ in verified stipend payments across 3 payment cycles. The 19-check integrity system catches data anomalies before they reach stakeholder reports. Field workers went from waiting days for task assignments to getting real-time updates. The World Bank gets automated compliance reports instead of manually compiled spreadsheets. As of 2025, the platform is scaling to serve 4 additional settlements across Nakuru and Kisumu counties.
50K+ / day
Daily API Requests
<200ms
Avg Response Time (ms)
76-table schema
Database Tables
KES 2M+
Payment Cycles Processed