Poor posture is a leading contributor to chronic back pain and musculoskeletal disorders, especially for individuals with sedentary lifestyles. Existing solutions within embedded systems often target clinical use and rely on expensive and extensive multi-sensor systems which limit the accessibility and affordability of such systems for individual users. This project investigates the design and implementation of an affordable, real-time wearable embedded system for posture monitoring and correction. The proposed system uses dual MPU-6050 IMU sensors, haptic and auditory feedback actuators, and Bluetooth Low Energy (BLE) to communicate with a mobile companion application. A server-based machine learning model trained on user data classifies posture types and qualities to enhance personalized posture correction. The embedded device was developed using Embedded C with the FreeRTOS real-time operating system, and the Raspberry Pi Pico Software Development Kit (SDK). The implementation of the project followed the Design Science Research (DSR) methodology, and was evaluated through a test protocol with controlled experiments. The resulting prototype system achieved a classification accuracy of 84% using a Random Forest machine learning model, and was kept under the aspired manufacturing cost target of less than 50 euros per unit. These results affirm the feasibility of the proposed system as an affordable and accessible solution for preventive healthcare and everyday ergonomic correction.