Wyza

Wyza helps companies create research studies, recruit and manage respondents, run AI-moderated interviews and turn responses into reports: AI-generated surveys, interviews in 40+ languages, transcription, emotion and persona analysis, and automated reporting that turns raw responses into a presentation-ready report in under 48 hours.

The platform is three Next.js apps sharing one monorepo — a research studio for brands, a respondent and field-agent app, and an admin terminal. I contributed ~1,270 of 2,220 commits (~57%) in a three-person engineering team.

Researchers struggled to choose survey types, configure targeting and set up conditional logic — so I helped redesign the core survey-creation flow around a ChatGPT-style experience that turns a research brief into a structured survey and recommends audience targeting. We tested with 20 researchers before rollout; measured survey-creation time dropped by up to 70%, and multi-hour setup flows became minutes.

When Customer.io became too expensive, I replaced it with an internal lead-management workflow — sales-pipeline controls, demo-request alerts and Slack notifications built into the admin product — removing the subscription entirely, an estimated $8k saved over 8 months.

Long-running AI and video work can’t live inside serverless limits, so I architected the asynchronous side: BullMQ/Redis and pg-boss queues feeding persistent Railway workers, with recovery and scheduled jobs. This is the infrastructure behind Wyza’s 24,000+ lifetime responses.

Payments run on dual rails — Stripe and Paystack — with webhook idempotency, a credits and wallet ledger, field-agent payouts, bank verification and KYC via BVN/Dojah.

I built /resolve-issue, a Claude-based workflow connected to Linear and GitHub through MCP. It has handled 150+ issues — from bug fixes to v1 features — automating much of the path from a written ticket to an opened PR, and gave the PM enough tooling to resolve lighter bugs without waiting on an engineer.

I also built Gandalf, an internal AI knowledge system used by all 18 team members: product work, sales context, tasks, meeting notes and project status in one place, automated team check-ins twice a day, and a daily project-health view for the CEO.