About me
Machine Learning Engineer at JPMorganChase, London. Ex-InstaDeeper who was building DeepPCB. Distinction in MSc Artificial Intelligence from Imperial College and First Class BSc in Mathematics from the University of Southampton.
Career
Developing AI-powered products that automate processes, prioritise tasks, and improve decision-making.
Responsibilities
- Build and deploy scalable Machine Learning services integrated with strategic systems.
- Communicate AI capabilities and results effectively to technical and non-technical audiences.
- Advance state-of-the-art AI applications in financial services, leveraging research in NLP, Computer Vision, and statistical machine learning.
- Maintain, monitor, and enhance existing machine learning models over time.
InstaDeep specializes in applying artificial intelligence (AI) to solving complex real-world problems. Primary examples of products impacted by my work are DeepPCB and DeepPack.
Responsibilities
Develop DeepPCB & DL-themed libraries.
- Develop in Python a Reinforcement Learning Distributed Library, using Ray, JAX, TensorFlow and DeepMind libraries.
- Implement in C++ an Inference Server for training large Neural Networks.
- Support Research and Applied teams with identifying and implementing solutions.
- Upskill members of the Research and Applied teams with respect to best Software Engineering practices.
Impact
Developed a Distributed Reinforcement Learning (RL) library as a member of the Research team working with the Product teams, using JAX, TensorFlow, Ray and DeepMind libraries. Accomplishments using Python and C++:
- Promoted as a result of achieving for all future experiments, 95% CPU and 39% GPU cost savings whilst reducing training time by 61%, for one of the applied revenue-making products. Primarily attained by implementing a distributed Asynchronous Proximal Policy Optimisation (APPO).
- Increased throughput by 321% and reduced all experiment costs by 60% by implementing a C++ gRPC server hosting GPU TensorFlow functions called by CPU only machines.
- Developed a multi-phase pipeline, starting with Behavioural Cloning, followed by Transfer Learning into Curriculum training, in partnership with one of the product teams.
- Wrote a Transformer in JAX and successfully trained it for one of the product teams.
- Used SEED-RL, a centralised inference server, for scaling the training of billion parameter models.
Levelise provides a platform for managing and optimising home batteries and solar panels in the UK and Australia.
Responsibilities
Develop RESTful backends and interfacing with home batteries.
- Write a Go backend to collect data from customer’s battery and PV systems.
- Write a Go backend for providing data to a battery system optimisation service.
- Write a Go backend for displaying system performance information to customers and customer support.
- Write Python Go code to identify and communicate system faults to installers and customer support staff.
Impact
Improved real-time monitoring and data aggregation of IoT devices used for controlling thousands of domestic batteries hosted across the UK and Australia for electricity grid stabilisation services. Accomplishments using Go and Python:
- Sped up by 10x the data aggregation of near real-time device per-second statistics.
- Achieved a 15x faster run-time for calculating daily statistics about their IoT devices.
- Analysed data and developed methods to automatically classify their customer IoT devices with fault codes and automatically notify the customer support team. I increased the range of faults that we could detect by 80%.
Education
Distinction
Modules: Deep Learning; Probabilistic Inference; Machine Learning for Imaging; Reinforcement Learning; Python Programming; Advanced Robotics; Symbolic AI; Ethics, Privacy and AI in Society; Group Project: Deep Learning on Brain Surfaces; Individual Project: Understanding Complex Flows - Classifying Spinning Vortices Using Machine Learning.
First-Class Honours
Modules: Optimisation; Mathematical Finance; Number Theory and Cryptography; Geometry and Topology; Numerical Methods; Statistical Distribution Theory; Mathematical Modelling; Group Theory; Relativity and Black Holes; Vector Calculus; Algorithms; Probability and Statistics; Partial Differential Equations; Linear Algebra I & II; Mathematical Modelling; Differential Equations; Calculus; Analysis; Maths and your Future; Mathematics for the Modern World; Individual Project: Introduction & Implementation of Supervised Machine Learning Techniques to Predict Heart Disease.
Publications
Clément Bonnet et al. (2024). “Jumanji: a Diverse Suite of Scalable Reinforcement Learning Environments in JAX”.
Submitted: NeurIPS 2023 Datasets and Benchmarks Track. doi:10.48550/arXiv.2306.09884
Vosylius, Wang, Waters et al. (2020). “Geometric Deep Learning for Post-Menstrual Age Prediction Based on the Neonatal White Matter Cortical Surface”.
In: Proceedings of UNSURE 2020, pp. 174–186. Lecture Notes in Computer Science, Springer. doi:10.1007/978-3-030-60365-6_17
Other Interests
- Newport Marathon 2023
- Wokingham Half Marathon 2022, 2023 & 2024
- Saundersfoot Triathlon 2019
- Rugby, played for Redingensians and Ranelagh School
- LAMDA Level 2, Distinction
- PADI Advanced Open Water Scuba Diver