AI facial assessment with 3D visualisation, built over a year-long engagement.

The problem

Qoves offers science-based facial assessments and skincare recommendations. The product needed to take a user's uploaded images, analyse skin and facial features, combine that with personal data and environmental factors like region, temperature, humidity and wind, and produce a personalised assessment and product recommendations in real time. It also needed to show users what it was seeing, which meant 3D facial visualisation in the browser rather than a static report.

What we built

Assessment pipeline. Image upload and processing, integration with the computer-vision models that analyse skin regions and facial features, and a recommendation engine that combines those results with the user's profile and live environmental data from weather and climate APIs.

3D visualisation. Three.js facial modelling so users see the regions being analysed and get a realistic preview, rather than a list of scores.

Guided workflow. A step-by-step assessment flow from upload to recommendations, built in React and Next.js with Tailwind.

Back end and infrastructure. Node.js and Express services, MongoDB and Redis, containerised with Docker on AWS.

The engagement began with a landing page and grew into a fourteen-month lead-engineering role across three contracts.

Results

A working assessment product with real-time recommendations and interactive 3D previews, and a client who came back twice: after Qoves, the same founder brought us in for Blackalgo. His words on the Qoves work: reliable, responsive, a strong understanding of Three.js, and "our go-to for getting difficult work done."