Procurement Decision Modeling with Generative AI: Optimizing Partner Selection Through Interactive Content Signals

FreeFuse is exploring how the choices users make within interactive content can reflect deeper preferences, risk tolerances, or learning gaps—which could also inform procurement or partnership decisions. In this project, students will design a procurement decision model that uses: User behavior data from FreeFuse content to infer buyer intent or partner alignment Generative AI tools (like ChatGPT or Claude) to simulate decision trade-offs (cost vs. risk, speed vs. quality) Scoring frameworks to evaluate which supplier or vendor path makes the most sense The final system should help recommend vendor or fulfillment options based on soft signals (user decisions), AI-driven inference, and traditional procurement factors.

Interactive Video-Controlled Retail Automation System

This project focuses on developing an interactive video system that enhances customer engagement in retail by allowing users to make real-time choices in a FreeFuse-powered interactive media experience. Instead of integrating a full IoT system, this project will simulate product selection and recommendation processes, helping students gain practical experience in interactive media design and decision-based user journeys. By simplifying the project scope, students will develop a robust understanding of interactive video technologies and how they can influence consumer behavior in retail settings.

Real-Time Object Interaction Layer for Live Video Streams

This project challenges students to develop a computer vision interaction engine that works on live video streams, not just pre-recorded content. The system will detect objects in real time using camera input (e.g., webcam or IP camera), and allow those objects to be tagged or interacted with by a viewer via a frontend UI. These tags will serve as triggers to launch new logic paths in the FreeFuse platform or send control messages to other systems (like IoT devices or content recommendation modules). Students will build: A live object detection pipeline using a webcam or stream input A frontend overlay interface that tracks objects in real time A logic module that turns user clicks/tags into structured events (e.g., “Object A → Launch Path B”) Student Learning Experience: Students will explore: Real-time computer vision using OpenCV and YOLOv8 Low-latency frame handling with WebRTC or MediaPipe Full-stack development with frontend overlays (Angular or React) API-based triggering logic using Node.js or Express Optional integration with downstream services (e.g., MQTT, FreeFuse logic router) Sponsor Support: Weekly check-ins Midpoint code review and prompt optimization feedback Access to example interaction flows from FreeFuse use cases

MQTT-Based IoT Logic Sync System

This project focuses on integrating FreeFuse’s interactive video decision engine with real-world IoT devices. Students will develop a middleware solution that listens for video-based interaction signals (e.g., a user selects a product or completes a task) and relays them to MQTT-connected devices, enabling live IoT responses to media engagement. The end system should: Use MQTT to publish structured interaction events Simulate or connect to edge devices (e.g., ESP32, Raspberry Pi) Sync device behavior with decision nodes in a multipath video journey This bridges the digital and physical, opening up use cases for training, kiosks, accessibility, and live response systems. Students will work across: MQTT communication protocols Angular frontend logic tying user actions to back-end device triggers Node.js middleware and message brokers Simulated or real device-side listeners (Python, C++) Basic edge security and QoS evaluation Sponsor Involvement: Weekly meetings with technical lead + midterm and final demos Documentation feedback + project planning support