International Journal of Science, Engineering and TechnologyAuthors: Indrajeet Trigunayat, Dr. Dharmbir Yadav
Abstract: Generative artificial intelligence has become an important capability for modern digital platforms, but production adoption requires more than model access. Enterprises need secure authentication, tenant isolation, policy-driven governance, auditability, observability, cost control, data protection and resilient operations. This paper presents the design and implementation of an enterprise-grade multi-tenant generative AI platform using FastAPI and Microsoft Azure. The proposed platform introduces a layered architecture consisting of a client access layer, API gateway layer, application orchestration layer, model routing layer, retrieval and knowledge layer, data persistence layer and observability/governance layer. FastAPI is used for high-performance asynchronous API development, dependency injection, request validation and modular service composition. Azure services are used for hosting, identity integration, secret management, monitoring, storage, database services and managed access to large language models. The paper defines a practical multi-tenancy model that separates tenant identity, policy, rate limits, usage budgets, data boundaries and audit events. It also proposes a secure request lifecycle for chat, retrieval-augmented generation and API-based consumption. The contribution of this study is a reference architecture and implementation methodology that can help academic and enterprise teams build scalable, secure and maintainable GenAI systems without tightly coupling application logic to a single model provider. The proposed design supports responsible AI controls, operational telemetry, disaster recovery readiness and future extensibility for additional models, connectors and agentic workloads.
DOI:
