Research

• ICETCE 2025 · IEEE Xplore · Feb 19, 2026Conference Paper

  HireIndex: A RAG-Enhanced AI Recruitment Platform for Fair, Transparent, and Efficient Hiring

  Abhishek Tiwari

  This paper presents a RAG-enhanced AI recruitment framework integrating DistilBERT-based semantic resume–job matching with SBERT and FAISS-based dense retrieval to incorporate contextual knowledge during inference. The proposed pipeline combines adversarial debiasing and equalized-odds post-processing for fairness-aware decision-making, along with SHAP-based explainability to generate transparent recruiter-facing rationales within a scalable end-to-end evaluation system.

  ICETCE 2025 · IEEEIEEE Xplore
• Indian Patent · Application No. 202611047414 A · Published 22/05/2026Patent

  RAG-Improved Artificial Intelligence Recruitment System, Fairness, and Explainability

  Dr. Neeraj Kumar Verma, Abhishek Tiwari, Adit Katiyar, Sneha Dhanuka

  An AI-powered recruitment platform comprising an integrated system of interconnected modules for automated candidate screening and selection. The system features a dual embedding architecture employing DistilBERT and Sentence-BERT, a retrieval-augmented generation layer with FAISS-based vector indexing, a multi-stage fairness enforcement mechanism, and a SHAP-based explainability module. Achieves ~98% screening accuracy — outperforming conventional ATS at 54% and standalone transformer models at 68–74% — while delivering a 72% reduction in recruitment bias.

  IN Patent App. 202611047414 A · Manipal University Jaipur
• Indian Patent · Application No. 202511056772 · Issued Jun 27, 2025Patent

  Vertically Integrated Data Centre System with AI-Based Cooling Along with PCM Based Immersion Cooling

  Abhishek Tiwari

  An advanced vertically integrated data centre system combining AI-based dynamic cooling and PCM-based immersion cooling to redefine thermal efficiency in high-performance computing. An ML-powered control layer monitors real-time heat maps, server load, and environmental parameters to autonomously regulate cooling intensity and airflow, optimising PUE and preventing thermal hotspots. Servers are immersed in a non-conductive liquid infused with Phase Change Materials that absorb heat during phase transition, reducing reliance on active cooling systems and enabling ultra-dense hardware configurations. The design supports compact stackable server modules, minimal physical footprint, reduced cooling overhead, high-density AI/ML workloads, and environmentally sustainable operation — targeting hyperscale, edge computing, and green AI infrastructure.

  IN Patent 202511056772 · Issued Jun 27, 2025