Papers

Papers I've written, contributed to, or frequently reference in my research.

Thesis
AI for Fetal Heart Ultrasound: MobileUNet, AutoFHR, and L-FUSION

Abhivir Singh · MEng Thesis, Imperial College London

Developing lightweight deep learning models for automated fetal heart analysis in ultrasound. Targeting deployment on portable devices for low-resource clinical settings.

Publication
MEGA: Multilingual Evaluation of Generative AI

Abhivir Singh et al. · 2024

Comprehensive benchmarking study evaluating large language models across 70+ languages on diverse NLP tasks.

Reference
Attention Is All You Need ↗

Vaswani et al. · NeurIPS 2017

The paper that started the transformer revolution. Still worth re-reading every few months.

Reference

Foundational architecture for medical image segmentation. Directly relevant to my thesis work on fetal ultrasound.

Reference

Depthwise separable convolutions for efficient inference. Key to deploying models on portable ultrasound devices.