← All workAI
PicsZako
An AI event-photo platform. Guests find their own photos with a selfie, and pay by M-Pesa.
- Client
- Wazomind
- Role
- Founder, product and build
- Year
- 2025, present
Not hosted
Runs locally. Not publicly hosted.
- money-handling product
- Livemoney-handling product
- payments with photographer payouts
- M-Pesapayments with photographer payouts
The problem
A photographer shoots two thousand frames at a wedding. Guests want the eleven they appear in. The manual answer is a WhatsApp group, a Drive folder and hours of admin, and no photographer gets paid for any of it. The global platforms that solve this do not take M-Pesa, which makes them unusable for the Kenyan market they would otherwise serve.
What I built
- 01Face-matching delivery, a guest uploads a selfie and gets only their own photos back, powered by a Python DeepFace service behind the Node API.
- 02Automatic watermarking on previews, with the clean file released on purchase.
- 03M-Pesa checkout with a photographer wallet and payout flow, so the money reaches the photographer without manual reconciliation.
- 04PayPal alongside M-Pesa for diaspora and international clients.
- 05WhatsApp marketing built into the delivery flow, because that is where Kenyan guests actually are.
- 06A full product and UX remap benchmarked against the category leaders, then a hardening pass across the live payment paths.
How it was measured
- Health and readiness probes on the API, with strict environment validation that refuses to boot a misconfigured production server.
- Background cleanup of uploaded selfies after a fixed retention window, a privacy requirement, enforced in code.
Stack
- Next.js
- Node / Express
- PostgreSQL
- Computer vision, DeepFace
- M-Pesa
- PayPal
Result
A live product taking real payments, with a Kenya-native feature set the category leaders do not offer. Built end to end, concept, UX, full-stack build, payments, deployment, and marketed by the same hands.