I'm a machine learning enthusiast interested in machine learning systems, robotics, and probabilistic modelling. Most of my projects explore how learning algorithms interact with real systems; simulations, data pipelines, and decision processes.

• Reinforcement learning for embodied systems
• Vision-Language-Action and world model architectures
• Probabilistic modelling and mixture density networks
• Machine learning infrastructure and experimentation pipelines

• Go2 Locomotion Policy: Reinforcement learning pipelines for Unitree Go2 quadruped locomotion in IsaacLab using JAX and PPO. Includes a custom ZeroMQ communication layer linking the simulator and training processes.

• CLIP-Flax: Library for importing CLIP weights from PyTorch into JAX/Flax NNX with numerical verification.

I’m particularly interested in problems that involve reasoning under uncertainty and building systems where learning algorithms interact with complex environments.

• Email: williams.edi012@gmail.com
• LinkedIn: https://linkedin.com/in/williams-edi