Project Mentorship & Technical Leadership: Imperial College CampusSphere
Duration: April - June 2025 | Collaboration: Imperial College London & Microsoft
Executive Summary
As a technical mentor and industry partner, I led a collaboration with Imperial College London to develop "CampusSphere," a pioneering multi-agent smart campus system. My role was to bridge the gap between academic AI research and enterprise-grade implementation, guiding a team of talented students in building a functional prototype using Microsoft's agentic frameworks and Azure AI services. The project successfully demonstrated the practical application of multi-agent systems in a real-world environment, culminating in a working prototype and a feature article on the Microsoft Tech Community blog.
My Role & Core Contributions
As the primary industry mentor, I was responsible for the strategic and technical direction of the project. My goal was to translate industry best practices into an effective academic project.
-
Technical Architecture & Strategy
I guided the design of the foundational multi-agent architecture, leading the team in decomposing complex campus functions into specialized, autonomous agents, and architected the inter-agent communication protocols.
-
Mentorship & Knowledge Transfer
I mentored the student team through the entire development lifecycle, teaching them enterprise-grade practices like scalable system design, production readiness, performance optimization, and security compliance.
-
Technology Implementation Guidance
I provided expert guidance on implementing the technology stack, including the creation of over 30 custom Model Context Protocol (MCP) tools and the use of Azure Cosmos DB for hybrid search.
Project Collaboration
A key aspect of this project's success was the strong partnership between the academic and industry mentors. The primary stakeholders included:
- Academic Supervisor: Abd Al Rahman Ebbayeh (Imperial College London)
- Industry Representative: Lee Stott (Microsoft)
My role was to provide direct, hands-on mentorship to the student team. I was responsible for overseeing their technical progress, equipping them with the necessary Microsoft resources and expertise, and resolving any blockers they encountered. I worked closely with the students to guide their development, ensuring their work remained aligned with the project's vision in coordination with the key stakeholders.
Technical Deep Dive
I guided the team in designing a scalable and robust system built on a modern, cloud-native stack. The core of CampusSphere is a team of specialized agents, each responsible for a distinct domain of campus management.
Agent Specializations
Function: Optimized building HVAC and lighting systems.
Implementation: Integrated with campus IoT sensors to monitor real-time energy consumption, analyze patterns, and autonomously adjust systems to reduce waste while maintaining comfort levels.
Function: Coordinated campus surveillance and access control.
Implementation: Monitored live video feeds and access logs to automate threat detection, manage access permissions, and trigger emergency response protocols when anomalies were detected.
Function: Implemented predictive maintenance for campus facilities.
Implementation: Analyzed equipment usage data and sensor readings to forecast potential failures, automatically generating and assigning work orders to the maintenance staff to minimize downtime.
Function: Provided an intelligent, 24/7 point of contact for students.
Implementation: Handled service requests and answered queries by performing semantic searches on campus knowledge bases (course info, event schedules, facility bookings) using Azure Cosmos DB's vector search capabilities.
The technology stack was chosen to provide an enterprise-grade, scalable, and secure foundation for the multi-agent system.
-
Azure AI Foundry & Agent Service
For creating, managing, and orchestrating the autonomous agents.
-
Azure Cosmos DB
Utilized for its powerful hybrid search, combining traditional SQL with semantic vector search for the agents' knowledge base.
-
Custom MCP Tooling
Developed over 30 custom Model Context Protocol (MCP) tools to allow agents to interact with real-world campus APIs and IoT systems.
-
Azure Production Infrastructure
Deployed with a focus on enterprise-grade security, performance optimization, and disaster recovery protocols.
Outcomes & Measured Impact
The mentorship resulted in significant achievements, delivering value to the students, the university, and the broader tech community.
-
Educational & Student Impact
- Real-World Experience: Students gained invaluable, hands-on experience in deploying sophisticated AI systems on Microsoft Azure.
- Bridged Theory & Practice: The project successfully translated theoretical AI concepts into a practical, working application.
- Powerful Portfolio Asset: The delivered prototype and public repository serve as a cornerstone of the students' professional portfolios.
-
Technical & Industry Impact
- Functional Prototype: Successfully delivered a working multi-agent system that can perform key campus management tasks autonomously.
- Microsoft Recognition: The project's innovation was recognized with a feature article on the official Microsoft Tech Community blog.
- Reference Architecture: The project now serves as a blueprint for other institutions looking to leverage agentic AI for campus modernization.
References & Media
-
Featured Article
Showcased on the Microsoft Tech Community blog as a model for academic-industry AI collaboration.
-
Project Repository
The complete source code, documentation, and agent architecture are available on GitHub.
-
Academic Institution
A collaboration with Imperial College London, a world-leading university in science, engineering, and medicine.