Vaibhav Dubey

National Institute of Technology, Kurukshetra

B.Tech · Electrical Engineering

August 1, 2022 – June 30, 2026

Zlice

AI Intern

January 1, 2026 – May 1, 2026

India

Hybrid AI Recruitment Platform

January 1, 2024 – June 1, 2026

• Architected a scalable AI recruitment engine for semantic candidate search, ranking, recommendation, and explainable hiring decisions across 100,000+ candidate profiles. • Engineered a hybrid retrieval and Learning-to-Rank pipeline that improved retrieval-signal detection from 21.6% to 43.0% while increasing ranking quality across enterprise hiring datasets. • Designed automated evaluation and quality auditing pipelines achieving 82% retrieval coverage, 66% ranking coverage, and 56% recommendation coverage within Top-100 candidate recommendations. • Optimized end-to-end candidate processing to rank 100K profiles in ~26 seconds while generating explainable recommendations and validated hiring reports.

Autonomous Enterprise Manager

January 1, 2024 – June 1, 2026

• Architected a modular Enterprise AI Operating System with layered knowledge, memory, orchestration, capability, governance, and observability architectures for scalable multi-agent enterprise automation. • Built a cognitive knowledge platform capable of ingesting enterprise documents and GitHub repositories, performing semantic retrieval, long-term memory reasoning, and grounded question answering with 94% grounded-answer rate, 89% citation accuracy, and 85% Recall@5. • Designed a three-layer cognitive memory architecture combining episodic, semantic, and structured memory with asynchronous extraction, importance scoring, and deduplication, achieving 90% memory precision, 92% deduplication efficiency, and 88% prompt token reduction. • Developed a benchmarking and evaluation platform for enterprise AI systems, validating multi-agent workflows over 50 benchmark iterations and achieving 64.7 ms retrieval latency, 2.23 s end-to-end latency, and 92.5% workflow success.

Early Prediction of Lithium-Ion Battery Remaining Useful Life

January 1, 2024 – June 1, 2026

• Developed an intelligent prognostics framework for early Remaining Useful Life (RUL) estimation using NASA battery degradation datasets and advanced feature engineering. • Engineered degradation-aware feature extraction pipelines capturing capacity fade, State-of-Health trends, and temporal degradation patterns for robust predictive modeling. • Built and benchmarked multiple machine learning and deep learning models, achieving a best Test MAE of 51.63 while extending the framework to LSTM and Transformer architectures for long-horizon prediction.