Michael Clark
Director at Cytophenix
Greater Perth Area Australia
Summary
🤩
Rockstar🎓
Top SchoolMichael Clark is a Director and machine learning engineer with 13 years of industry experience and 8+ years focused on ML and AI alignment, currently helping spin out medical AI at Cytophenix while acting as a Deep Learning SME for Woodside Energy. He blends hands-on model and systems work—contributions to notable open-source projects like Keras-contrib, Tensorforce and recurrentshop show expertise in loss functions, prioritized replay and robust recurrent architectures—with product delivery across energy, mining and geospatial domains. As founder/consultant through Three Springs and ThinkCDS he has shipped forecasting, anomaly detection and satellite-imagery solutions, and taught subsurface ML via an open curriculum. He brings a geophysics and physics background (BSc Hons, MSc Petroleum Geoscience) that helps bridge domain data challenges and ML engineering. Colleagues value his practical focus on robustness (NaN fixes, unit tests, replay-buffer persistence) and a wry sense of terminology hygiene—if you can’t remember what something does, it’s a “wassname.”
13 years of coding experience
5 years of employment as a software developer
BSc (Hons), Physics, First Class, BSc (Hons), Physics, First Class at University of CanterburyMaster of Science (MSc), Petroleum Geoscience, Master of Science (MSc), Petroleum Geoscience at Victoria University of Wellington
English
Stackoverflow
569reputation
433kreached
6answers
0questions
Github Skills (37)
physics-engine10
pytorch10
debug10
algorithms10
phaser10
convolutional-neural-networks10
javascript10
python10
pandas10
machine-learning10
recurrent-neural-networks10
reinforcement-learning10
numpy10
keras10
deep-learning10
Programming languages (14)
Github contributions (5)
wassname/rl-portfolio-management
Jul 2017 - May 2019
Attempting to replicate "A Deep Reinforcement Learning Framework for the Financial Portfolio Management Problem" https://arxiv.org/abs/1706.10059 (and an openai gym environment)
Role in this project:
ML Engineer
ML EngineerContributions:151 commits, 2 PRs, 77 pushes in 1 year 10 months
Contributions summary:Michael's commits focused on implementing and testing utility functions related to data processing within a deep reinforcement learning project for portfolio management. They implemented random shift, normalization, and scaling functions, which were then tested within the environment, ensuring the correct behaviour of data preparation steps. Further contributions involved improving the environment's functionality, including adding a holding cost.
financialreplicatearxivportfolio-managementabs
ShangtongZhang/DeepRL
Nov 2017 - Jun 2018
Modularized Implementation of Deep RL Algorithms in PyTorch
Role in this project:ML Engineer
ML EngineerContributions:8 commits, 7 PRs, 5 comments in 7 months
Contributions summary:Michael primarily contributed to improving the robustness and stability of deep reinforcement learning algorithms implemented in PyTorch. They addressed NaN issues by adding epsilon values and modifying code related to the calculation of standard deviations and log densities. The user also implemented saving and loading functionality for the replay buffer and normalizers. Furthermore, the user made code changes related to PPO to work with GPU and updated dependencies.
pytorchdeep-learningreinforcement-learningmachine-learningmodularized
Find and Hire Top DevelopersWe’ve analyzed the programming source code of over 60 million software developers on GitHub and scored them by 50,000 skills. Sign-up on Prog,AI to search for software developers.