João GabrielSoftware engineer
Backend engineering & applied AI
All work

Applied AI · Academic prototype

GymIA

An academic workout-planning prototype that helps qualified fitness professionals generate, edit, and manage structured training plans.

Scope
Full-stack engineering & AI workflow
Built with
React, Express, Prisma, OpenAI
GymIA interface for organizing and reviewing a client’s workout plan.
Applied AI · Academic prototype / Academic prototype
Open full-size image

The problem

Personal trainers balance client histories, activity, and the details of every workout. GymIA explores where language models can reduce the drafting work while leaving professional judgment in control.

An AI-generated plan starts the conversation. The instructor can review exercises, sets, repetitions, intervals, and load in a structured editor before exporting a sheet for the client.

Engineering decisions

01

Keep the result editable

The model produces a structured workout plan. A human review step turns it into an instructor-owned artifact instead of an autonomous prescription.

02

Model the real relationships

Express, Prisma, and MySQL organize instructors, clients, plans, and chat sessions. JWT authentication scopes access to the workspace.

03

Finish the practical workflow

Client activity views, plan editing, and print-friendly export bring the prototype beyond a standalone AI chat box.

What shipped

An academic full-stack prototype with client management, AI-assisted plan generation, an instructor review interface, and PDF and print export. Generated suggestions require review by a qualified professional.

Read the source
Next projectPensieve