Abhivir Singh

AI & ML at Imperial College London

Final-year MEng Computing (AI & ML) student at Imperial. I work on deep learning for medical imaging, build ML systems, and occasionally trade derivatives. Currently writing my thesis on AI-guided fetal ultrasound analysis.

I'm studying at Imperial College London, having transferred from IIT Delhi after my first year. My focus is on machine learning research — particularly deep learning for medical applications and natural language processing.

Before university, I was a KVPY Fellow, qualified the Regional Math Olympiad, and scored 9.0/9.0 on the ENGAA. I've received multiple Gold Medals for academic performance. Outside of research, I'm interested in quantitative finance, systems programming, and building things that work well.

See all work →

AI for Fetal Heart Ultrasound

2024 — Present

Ongoing MEng thesis at Imperial College London. Developing MobileUNet for real-time fetal heart segmentation, AutoFHR for automated heart rate estimation, and L-FUSION for landmark-guided probe navigation. Targeting deployment on portable ultrasound devices for low-resource settings.

Deep LearningMedical ImagingPyTorchUltrasound

Automated ML Platform

Summer 2025

Built an end-to-end automated machine learning platform at Fetch.ai using LangChain agents for intelligent pipeline orchestration. Implemented stacked ensemble methods with automated hyperparameter tuning, feature engineering, and model selection.

LangChainAutoMLPythonEnsembles

AI-Powered Text-to-Speech for Ads

Winter 2024/25

Developed production voice synthesis at DeepSearch Labs using MaskCycleGAN-TTS for voice conversion and VITS for end-to-end speech synthesis. Enabled personalised ad audio generation at scale.

TTSGANsVITSSpeech Synthesis

GPT-2 from Scratch

Summer 2024

Implemented GPT-2 (124M parameters) from scratch as part of Eureka Labs. Built the full transformer decoder architecture including multi-head self-attention, positional encoding, and byte-pair encoding tokenizer. Trained on OpenWebText.

TransformersNLPPyTorchLLMs
paper
article
book
Designing Data-Intensive Applications

Martin Kleppmann