Dévorer: Redefining the Campus Dining Experience with AI
How an operator of next-gen catering engaged a full-stack engineer, then a full product team, then two mobile engineers.
Challenge
Dévorer puts global street food into campuses, stadiums, offices and hospitals, rotating more than four hundred vendors and forty cuisines across its sites. The part that makes the model work is not the food, it is the decision behind it: which vendor goes to which site, on which day, in what quantity. Demand moves with term dates, fixtures, weather and the hour. That is a forecasting, scheduling and analytics problem at the scale of hundreds of vendors and dozens of locations. It is a software company growing inside a food company.
The engagement started deliberately small: one full-stack engineer on React, Node.js, GraphQL and AWS, chosen as much for the shape of his previous work as for the stack. Once that stage showed what the platform could be, the demand widened to a team around him: a backend engineer who had built multi-tenant logistics orchestration and a live assignment engine on AWS, an AI engineer with production RAG pipelines and fine-tuned models, a data analyst who designs layered warehouses and automates them in Airflow, a DevOps engineer on Terraform and Kubernetes, a React frontend engineer, a QA engineer on Playwright and Postman, and a project manager accountable for delivery. The third stage went to mobile: two Flutter engineers, one a technical lead with eight years of cross-platform delivery, the other working in offline-first architecture and native integrations.
Each stage was screened against the problem actually in front of it rather than against a standing specification, which is why the backend and mobile shortlists looked nothing alike. By the third stage Dévorer had hired a CTO, who ran that round of interviews himself, while the founders had run the first two. The opening match was finalised in 48 hours after the first request, and the engineer placed then was still on the project three stages later.
Outcome
Dévorer team did not try to hire a department. They hired one engineer, found out what was possible, and then brought in the team the answer required. By the time the work reached mobile, they had a CTO in post to run it. Three stages across ten months of engagements, each specified by what the previous one had established.
- Ten requests across three stages, the first agreed inside 48 hours.
- A complete product organisation: platform, backend, data, AI, infrastructure, frontend, testing, delivery management and mobile.
- Built around two non-technical founders, starting with a single engineer and growing into a full team before a CTO joined to run the final stage.
- Ten months of engagement in with the product live on the web and on both app stores.
Working with Yotewo has been a pleasure. I’d like to specifically highlight Aliaksandr, the founder, who is always extremely prompt in his responses and highly attentive to requests. Yotewo is tackling a real problem, and I would gladly recommend them.

A Strategic, Multi-Phase Hiring Approach
Phase: Request
Requests in three stages rather than one: a single full-stack engineer to establish what the platform could do, then a full team around him once that was known, then two Flutter engineers to take it onto mobiles.
Phase: Matching
Each stage was screened against the problem in front of it: real-time systems tracking things across locations for the first engineer, multi-tenant orchestration and assignment engines for the backend role, production RAG rather than prototypes for the AI role, warehouse design and Airflow for data, and cross-platform applications shipped on both stores for mobile.
Phase: Interviews
The founders ran the first two stages themselves, and the CTO they had hired by then ran the third, with the opening engagement confirmed 48 hours after the request.
Phase: Engagement
Every engagement opened with a mix of full-time and part-time monthly workload, hour caps set before the start, visible daily hour tracking, and monthly reporting for hour commitments.
Phase: Support
The engagement grew and workloads changed month-to-month, with hour caps changes and extra hours being requested and approved by the founding team.
