Fault-Tolerant AI-Integrated HRMS Using Hybrid Generative and Deterministic Intelligence for Reliable Workforce Management
Fault-Tolerant AI-Integrated HRMS Using Hybrid Generative and Deterministic Intelligence for Reliable Workforce Management
Authors:
1st R. Ramakrishnan
Associate professor
Department of MCA
Sri Manakula Vinayagar Engineering College (Autonomous)
Puducherry- INDIA
ramakrishnanramasubramaniam@gmail.com
2nd S.Lalitha
PG-Student
Department of MCA
Sri Manakula Vinayagar Engineering College (Autonomous)
Puducherry- INDIA
Abstract
We present a hybrid architecture for intelligent human resource management that integrates a generative large language model with a deterministic fallback engine to achieve fault-tolerant digital workforce operations. The proposed system processes user queries through two parallel pathways: a primary probabilistic layer based on a decoder-only transformer model generates context-aware responses, while a secondary deterministic layer employs TF-IDF weighting and syntactic dependency parsing for intent classification against predefined templates. A dynamic routing function governs the final output selection by monitoring the availability of the large language model service and evaluating a confidence threshold on the generated response. If the generative service is accessible and its confidence score meets the threshold, the system delivers the transformer-based output; otherwise, it seamlessly substitutes the deterministic result from the fallback engine without requiring user re-query or manual intervention. This design effectively isolates the stochastic nature of generative artificial intelligence from critical human resource operations, thereby ensuring uninterrupted service continuity. The novelty lies in the explicit separation of probabilistic and deterministic inference paths, combined with a real-time substitution logic that maintains system reliability. Our approach significantly reduces the risk of service disruption in enterprise environments where generative model failures or timeouts could otherwise halt essential workforce management tasks.
Keywords: Human Resource Management System (HRMS); Artificial Intelligence; Fault-Tolerant Architecture; Large Language Models (LLMs); Natural Language Processing (NLP); Hybrid AI Systems; Conversational AI; Microservice Architecture.