Titulo Estágio
Design and Implementation of an AI-Driven Personalization System for Fitness Applications
Áreas de especialidade
Engenharia de Software
Local do Estágio
Coimbra
Enquadramento
Fitness apps are everywhere - but most still rely on fixed workout plans that don’t evolve with the user.
Whether you're recovering from a hard session, have limited time to train, or simply want smarter
recommendations, most apps offer little adaptability.
This project aims to change that.
We propose the development of a next-generation fitness app that uses artificial intelligence to tailor
workouts to each user's routine, recovery needs, and time constraints. The goal is to move away from
rigid plans and towards intelligent, responsive systems — just like Spotify recommends songs or Netflix
suggests shows, this app will suggest the right workout for the day.
From a technical perspective, the project combines:
Mobile development (user interfaces, session tracking, progress visualization)
Backend/API engineering (data persistence, AI integration)
Applied machine learning (recommendation systems, rule-based logic, ML models)
It’s a multidisciplinary challenge, ideal for students s
Objetivo
The main goal of this project is to design and implement a modular, AI-enhanced fitness application
that brings together mobile development, backend engineering, and applied machine learning. The
core objectives are:
1. Development of a fully functional mobile fitness app, including:
• Support for creating custom exercises (e.g., name, category, equipment, target muscles)
and assembling structured workout plans.
• Tracking of workout sessions, capturing detailed data such as sets, reps, weight, time
under tension, and rest intervals.
• Real-time session logic (e.g., dynamic timers, active set progression, automatic rest
logging).
• Visualization of progress through interactive charts (e.g., volume trends by muscle group,
personal records, training consistency).
2. Integration of AI-based features, including:
• A workout recommendation engine trained on historical workout logs and user
constraints (e.g., available time, muscle group fatigue), using rule-based filtering and
machine learning models (e.g., collaborative filtering, nearest-neighbor methods).
• Recovery guidance based on training load and frequency, using heuristics or predictive
models to detect undertraining and overtraining patterns.
• (Optional/Advanced) Pose estimation for form analysis, leveraging pre-trained models
(e.g., MediaPipe or OpenPose) to provide feedback on exercise execution.
3. Technical exploration and tooling:
• Freedom to select the mobile development stack (e.g., Jetpack Compose, Swi
Plano de Trabalhos - Semestre 1
• Research fitness
• Research technologies to
use
• Research AI approaches
• Define architecture
• Define DB schemas
• Define AI pipeline
Plano de Trabalhos - Semestre 2
Start with simulated data
• Build rule-based system
Train/refine models
• Integrate API
Run user testing
• Analyze AI
recommendations
Analyze AI
recommendations
Condições
Bolsa de Estágio
Observações
n/a
Orientador
Diogo Miguel Viana Maya Monteiro
diogo.m.monteiro@accenture.com 📩