Jillur Quddus is a computational mathematician, polyglot software engineer and published author who specialises in the design of novel mathematical models and the engineering of innovative, high-performance, ethical and secure artificial intelligence (AI) systems. Jillur is also the founder of HyperLearning AI which he started with the aim of ensuring that AI delivers tangible and positive social impact.
Jillur's core areas of expertise include graph theory and graph neural networks (GNNs) which he has applied to the design of real-time neural network software interfaces for embedded smart devices, emerging infectious disease modelling, preventative healthcare, and combatting Dark Web-enabled crime. Jillur has deep experience of working within central government, healthcare and law enforcement, and has worked extensively across the world including in Japan, Singapore, Hong Kong, Malaysia, Australia, New Zealand and the United Kingdom.
Core Skills
- Mathematical Modelling
Computational Complexity Theory • Number Theory • Graph Theory • Statistical Learning • Machine Learning • Deep Learning • Graph Neural Networks (GNNs) - Software Engineering
Python • Java • JavaScript • SQL • Gremlin • Shell Scripting - Distributed Computing and ML Frameworks
Apache Spark • Dask • PyTorch • TensorFlow • Apache TinkerPop • Elasticsearch
Selected Experience
- Joint Biosecurity Centre - Lead Data Scientist
Jillur led an expert team of globally leading mathematicians, epidemiologists, AI research fellows and technologists, responsible for developing bleeding-edge emerging infectious disease modelling systems. - Government (UK) - Lead Artificial Intelligence Engineer
Jillur led an expert team of intelligence officers, cryptographers, analysts and subject matter experts responsible for combatting Dark Web-enabled child sexual exploitation & abuse (CSEA) through the analysis of anonymous decentralised networks and cryptographic protocols. - Local Government (UK) - Lead Data Scientist & Software Engineer
Jillur designed and built an award-winning artificial intelligence system capable of autonomously detecting, classifying and prioritising cases of potholes and fly-tipping in real-time through the development of a bespoke deep convolutional neural architecture and neural network software interface capable of processing images at 45fps with less than 25ms latency delivering up to 40 detection events per second, integrated with a bespoke Android mobile app and GPS location services.
Publications and Books
- Machine Learning with Apache Spark
Uncover patterns, derive actionable insights and learn from big data using MLlib. Amazon • Waterstones • Google Play
Languages
- English - Native
- Japanese 日本語 - Advanced (日本語能力試験 JLPT N2)
- Chinese Cantonese 廣東話 - Intermediate