Author Affiliations
[5] Assistant Professor, Department of Computer Science and Engineering Er. Perumal Manimekalai College of Engineering, Hosur
[1] [2] [3] [4] Department of CSE, Er. Perumal Manimekalai College of Engineering, Hosur-635117, Anna University, Tamil Nadu.
Abstract
Enforcing strict data isolation between cohosted tenants while delivering intelligent workflow automation represents a core engineering challenge in enterprise Software-as-a-Service (SaaS) platforms. Existing issue-tracking systems typically enforce isolation at the application layer, where omitted filter predicates can silently expose cross-tenant data, or employ resourceintensive per-tenant schema partitioning. This paper presents SaaS Tickets, a production-grade multi-tenant issue-tracking platform that addresses these challenges through three primary contributions: (i) an AspectOriented Programming (AOP) interceptor that systematically activates Hibernate’s persistent @Filter on every repository method invocation, providing ORM-level SQL tenant scoping that is designed to prevent bypass through normal application code paths; (ii) a five-feature Google Gemini LLM integration pipeline employing asynchronous thread-pool decoupling to remove AI latency from the critical user-request path; and (iii) a real-time voice assistant architecture using Vapi.ai WebRTC transport with a custom OpenAI-compatible streaming LLM endpoint backed by Gemini, enabling hands-free ticket intelligence queries. Security evaluation across eight adversarial test scenarios confirmed correct enforcement of isolation, JWT validation, role-based access control, and graceful AI degradation in all cases. Classification experiments on 150 labelled tickets achieved a weighted F1-score of 0.86 across five category classes. Asynchronous ticket categorization reduced userperceived AI latency by a factor of 7.8 relative to synchronous processing. The fully containerised, clouddeployed architecture demonstrates viability as a scalable enterprise SaaS foundation.
Index Terms—Multi-Tenant SaaS, Hibernate Filter, Aspect-Oriented Programming, ORM-Level Tenant Isolation, Role-Based Access Control, LLM Ticket Classification, Google Gemini, Asynchronous AI Integration, Voice AI, Vapi, OAuth 2.0, Spring Boot, Docker.
How to Cite This Article
Dhinakar R, Karthick K, Keerthik Raj JP, Gokula Krishnan R, K. Egneswari (2026). SaaS Tickets: A Secure Multi-Tenant AIAugmented Issue Tracking Platform with ORMLayer Tenant Isolation and Gemini-Powered Workflow Automation. International Journal of Innovative Research in Multidisciplinary Education & Technology (IJIRMET), 11(3).