An Agentic System for Schema-Aware NL2SQL Generation
CAI 2026 Conference
Multi-agent architecture achieving 50.87% execution accuracy on BIRD benchmark while reducing inference cost by 90%, using locally finetuned SLMs through LoRA & QLoRA.
Publications
Research papers and academic contributions
Evaluating Credit Risk Using Interpretable Machine Learning Models
CADSCOM 2024 & AMLDS 2025
Interpretable credit risk models using Ensemble Models + LIME achieving 76% accuracy while ensuring regulatory compliance and full model transparency.