Forma: Building Personalised Workout Plans Around User Lifestyle
How an AI personal training app engaged a QA Team Lead, a QA Engineer and a Full-Stack Engineer.
Challenge
Forma is an AI personal trainer that rebuilds your week around whatever actually happened in it: a missed session, a short lunch break, a bad night’s sleep picked up from a watch. That adaptability is the product, and it is also what makes the product hard to verify. Every user gets a different plan, delivered across a web app and iOS, so there is no single correct screen to check against on either surface. Forma is a consumer subscription rather than enterprise software, which changes what a bug means.
Three requests went in together. Two were QA roles that shared a toolset but not a job: a Team Lead to establish how testing would work at all, because Forma had no QA function, and a QA Engineer to run it day to day. Both needed manual testing depth across web and mobile, end-to-end coverage including payment integrations and multi-language releases, and working practice in Postman, JMeter, TestRail and Jira inside a SCRUM team, with evidence of cutting a regression cycle rather than simply running one. The third was a Senior Full-stack Engineer with seven years and more across React, Next.js and TypeScript on the front end and Node.js and NestJS behind it, with GraphQL and REST, and a modernisation record to match: an AngularJS to React migration, a React 16 to 18 upgrade, and an NX monorepo that shortened release cycles across several teams.
The two sets of requests were matched against different evidence. For the testing roles it was regression time actually reduced on a live product, because testing a personalised output is a design problem before it is a workload. For the engineering role it was migrations completed in production rather than proposed, since modernising a product people use daily is the version of that work with no quiet weekend to do it in. The first engagement was confirmed 40 hours after the request opened, and Forma’s CTO and one of its founders ran the interviews and made the decisions between them.
Outcome
Releasing faster and testing properly are not two decisions. Forma needed both in the same week: an engineer whose record is in modernisation and shorter release cycles, and the two people who would build a testing function from nothing, so that a shorter release cycle was survivable on a product people open every morning.
- Three requests opened together across testing and engineering, with the first match confirmed in 40 hours.
- A QA function built from nothing, alongside the engineering capacity to modernise the product, it would have to cover and fix identified gaps.
- The engineer and both testers working directly with the CTO and a founder, inside the existing team rather than as a separate unit.
- Seven months into engagement covering release testing, migration work and day-to-day product development.
The team is always prompt with responses, happy to guide you through the platform, and very open to feedback. A great platform for non-technical founders looking to find and work with developers efficiently.

A Strategic, Multi-Phase Hiring Approach
Phase: Request
Three requests opened together rather than sequenced: a QA Team Lead to establish a testing function that did not yet exist, a QA Engineer to run it, and a full-stack engineer on React, Next.js, Node.js and NestJS, because the modernisation work and the testing that would have to cover it as well as fixing identified gaps were one decision.
Phase: Matching
Testing candidates were screened on regression cycles actually reduced on live products rather than test suites written, and the engineering candidate on migrations completed in production, with 5 of 9 shortlisted going through to interview.
Phase: Interviews
Forma’s CTO and one of its founders interviewed together and decided between them, with the first engagement being confirmed 40 hours after the request opened.
Phase: Engagement
Two of engagements opened with a part-time, and one with a full-time monthly workload, and hour cap set against the release cycle rather than a project deadline, with daily hour commitments in the tracker and monthly reports approved by CTO and founders.
Phase: Support
As the release schedule changed across seven months, workloads moved with it, and every change, such as extra hours, was requested and approved in the account.
