Tom Lemelle

With Milesia, a B2B2C platform connecting large retailers with their customers through AI.

With my two partners (Ugo Delalonde & Dorian Monaco), we created a simple and personalized shopping experience: the customer scans a QR code in-store or searches for a store online and answers a few questions (occasions, tastes, dishes, budget, colors, regions…) so that our AI recommends the three wines from the catalog that best match their taste.

Our approach combines quality and precision, each recommendation is based on real food-and-wine pairing rules, intelligent filtering (budget, color, region), before calling on the AI, which significantly reduces token consumption and speeds up data casting and recommendations.

On the technical side, I use the following stack:

Frontend in Next.js 15, React 19, and Tailwind CSS

Backend in AdonisJS 6, PostgreSQL, and Redis

DevOps using Docker with cache and monitoring on an OVH VPS.

For large retailers, Somelia offers a complete dashboard: tracking scans, product views, conversion rates, stock management, automatic alerts, and CSV/PDF exports.

I am passionate about creating useful and elegant tech products, where AI is not a gadget, but a real lever for experience and performance.

💻 As a freelance web developer, I have over 6 years of experience in the field, specializing in AdonisJS, Next.js (TypeScript), React, Vue.js, and Webflow to create impactful landing pages. I implement modern and tailor-made web solutions, adapted to each business.

🎥 Passionate about imagery, I also assist my clients in their photo and video projects to enhance their brand and universe in a coherent and creative way.

With the SafeEat project, we sought to bring the necessary simplicity to restaurateurs, canteen managers, and caterers by balancing food safety and peace of mind for their customers.

In this vein, we implemented a real-time platform to create recipes, organize menus, and be transparent with customers by displaying allergens via a dynamic QR code scannable from the table.

The user-facing part, on the other hand, is fluid and secure: each customer filters dishes according to their allergens and dietary restrictions to get real-time information on their mobile.

As CTO, I designed and supervised the entire technical and infrastructure aspect:

Stack: React, AdonisJS, TypeScript, PostGreSQL, Docker

CI/CD: GitHub Actions, deployment on Hostinger KVM 2 VPS

The project was successfully tested at the restaurant Le Bouillon d’Or (Rouen), proving a real potential for adoption in the restaurant industry.