The science behind your co-scientist
Nadhi isn't a wrapper. The methods it relies on, confidence-aware routing to curb hallucination, retrieval-augmented generation, reinforcement-learning agents, and multimodal medical interpretation, are our own peer-reviewed and preprint research. Our ECG interpretation paper is peer-reviewed in BMJ Digital Health & AI, and every paper below is on arXiv; read the originals and judge for yourself.
Full arXiv author profileHigh-accuracy ECG image interpretation using parameter-efficient Low-Rank Adaptation (LoRA) fine-tuning with multimodal LLaMA V.3.2
Nandakishor Mukkunnoth, Anjali Mukkunnoth, Meenakshi Khapre, Sudarshan Benamanahalli Puttaswamy
LoRA fine-tuning of multimodal LLaMA V.3.2 on one million annotated ECG images: AUC 0.98 against 0.51 for the base model, Macro F1 0.74 against 0.33. The research behind AI4Cardio.
Confidence-Aware Routing for Large Language Model Reliability Enhancement: A Multi-Signal Approach to Pre-Generation Hallucination Mitigation
Nandakishor M
Pre-generation hallucination mitigation via multi-signal confidence routing, directly underpins Nadhi’s “every claim traceable to a source” grounding.
SalesRLAgent: A Reinforcement Learning Approach for Real-Time Sales Conversion Prediction and Optimization
Nandakishor M
RL for real-time conversion prediction/optimization, the lineage behind Nadhi’s on-device SAC GTM forecasting.
ForcePose: A Deep Learning Approach for Force Calculation Based on Action Recognition Using MediaPipe Pose Estimation Combined with Object Detection
Nandakishor M, Vrinda Govind V, Anuradha Puthalath, Anzy L, Swathi P S, Aswathi R, Devaprabha A R, Varsha Raj, Midhuna Krishnan K, Akhila Anilkumar T V, Yamuna P V
Vision-based force estimation from pose + object detection.
DeepRAG: Building a Custom Hindi Embedding Model for Retrieval Augmented Generation from Scratch
Nandakishor M
A from-scratch embedding model for RAG, the same retrieval-augmented foundation Nadhi uses to synthesise your corpus.
Continuous Learning Conversational AI: A Personalized Agent Framework via A2C Reinforcement Learning
Nandakishor M, Anjali M
A personalized continually-learning agent framework via A2C reinforcement learning.
Listing reflects the team's arXiv record (author id mukkunnoth_n_1). arXiv preprints may not yet be peer-reviewed; where a paper has a journal version we link the authoritative source. We list our own work only, no third-party claims.