Orchestrix

Academic research groups face severe administrative bottlenecks and data vulnerability risks. Teams traditionally manage expensive shared laboratory assets (high-performance workstations, specialized sensors, GPU clusters, and testing rooms) using messy, fragmented spreadsheets, leading to frequent double-bookings and idle hardware. Furthermore, researchers frequently coordinate sensitive, pre-publication research over public cloud collaboration tools (like Slack or Trello), exposing proprietary intellectual property to third-party data breaches and AI model training scrapers.

Orchestrix solves this by providing a containerized Private SaaS platform deployed directly on isolated local lab infrastructure. Built on a decoupled polyglot microservices architecture (Spring Boot Core API, Python LangChain AI Engine, Node.js/Spring WebSockets, and PostgreSQL), it unifies interactive Kanban task tracking, real-time encrypted project chat, and an atomic database-locking algorithm for conflict-free lab resource scheduling. A localized background AI Context Engine automatically converts chat threads and meeting minutes into structured action items and suggested deadlines without transmitting data to external public AI APIs.

System Architecture & Design Methodology: Orchestrix is engineered across four distinct layers to ensure fault tolerance, strict data isolation, and low-latency interaction:

1. Presentation Layer: Built with React.js (Web Dashboard) and React Native (Mobile Companion), delivering intuitive Kanban boards, resource scheduling calendars, and rich-text collaborative editors.

2. Application Services Layer: Spring Boot microservices managing JWT-based authentication, Role-Based Access Control (RBAC across Institute Admins, PIs, Researchers, and Resource Managers), and dynamic schema-per-tenant database routing.

3. AI Context Service Layer: An independent Python/FastAPI service utilizing LangChain to parse natural language project communications in the background using localized models.

4. Persistence Layer: Single-tenant PostgreSQL database containerized via Docker, utilizing isolated PostgreSQL schemas per research organization (org_{slug}) to guarantee physical and logical data sovereignty.

Key Technical Achievements & Challenges Solved:

• Atomic Resource-Locking Algorithm: Prevents race conditions and overlapping equipment reservations during concurrent booking attempts by enforcing database-level transaction locks.

• Zero-Leak Multi-Tenancy: Custom Hibernate TenantIdentifierResolver and SchemaMultiTenantConnectionProvider dynamically set PostgreSQL search paths per request, ensuring zero cross-tenant data leakage.

• Privacy-Preserving AI Engine: Operates completely on local hardware, processing project documentation and chat threads to extract action items without risking IP exposure to commercial LLM APIs.

• Self-Hostable Packaging: Fully containerized with Docker Compose for single-command deployment on laboratory servers, requiring minimal sysadmin overhead.

Future Work: Planned enhancements include push notification services for mobile devices, MinIO integration for large experimental dataset uploads, and financial grant budget tracking modules.