What I learned building it.

Architecture decisions, product teardowns and hard-won patterns from shipping production systems, written for the engineers and product leaders who have to make the same calls. No theory I have not run in production.

7 publications Shipped, not theorised
Industry Report

The AI ROI number nobody had to verify

Fewer than a quarter of CEOs who invested in AI report revenue increases, and even that figure is executives grading their own work. PwC's AI performance study is well run, but the…

Applied AIMeasurementEnterprise AIEmerging MarketsROI
Case Study

HTTP 200 is not verification

One navigation link on this site returned 200 for an extended period while serving the wrong page entirely, and every status check we had passed the whole time. Three more defects…

TestingWeb PlatformAccessibilityData IntegrityEngineering Practice
Case Study

What the Free Tier Actually Costs

Rippli spent a year climbing down the managed-hosting ladder, Replit, then Render's free tier, then a bare VPS, because the workload was the wrong shape for free infrastructure,…

InfrastructurePlatform EngineeringGraph DatabasesCost
Whitepaper

Designing for Intermittent Infrastructure

Most apps assume the network is there. Across the markets I build for, it usually is not. Here is the architecture I keep coming back to, local-first writes, a durable intent…

Product ArchitectureEmerging MarketsOffline-First
Industry Report

Pay-to-play or pay-to-die: The coming AI subscription trap African CEOs must escape

Uber burnt through its entire 2026 AI budget by April, undone by a single engineer running up $1,200 in token fees in one debugging session. As Salesforce, Adobe and HubSpot…

AI EconomicsVendor Lock-InEnterprise AIEmerging MarketsInfrastructure Independence
Whitepaper

When Written Fluency Stopped Being a Signal

Fellowships, scholarships and grant programmes screen at scale on written English. Generative tools made that filter measure access rather than readiness, and detection cannot fix…

Selection IntegrityGoverned AIAssessment DesignProgramme DesignApplied AI
Whitepaper

Counting graduates is not measuring leaders

Selection programmes report how many people they admitted and graduated, and the ones that try to do better mostly reach for a bigger number. The opposite of a bad number is not a…

MeasurementLeadership DevelopmentEvaluationQualitative ResearchGovernance

No publications in this category yet.

Need this applied to your product?

I work with product and engineering teams on architecture, technical strategy and delivery, turning what is on this page into decisions, roadmaps and shipped systems.