My work focuses on Deep Reinforcement Learning, ML systems and Edge AI.
-
PPO-Belief: Researching a model-based extension of Proximal Policy Optimization by introducing an auxiliary transition prediction objective. The goal is to learn latent representations that improve decision making under partially observable and non-stationary environments. (Paper drafting in progress).
-
Kairos: Model-based RL architecture (PPO-Belief) predicting transition dynamics in chaotic financial environments. Built with PyTorch, WandB, Optuna.
-
Edge PaaS: Building an end-to-end deployment pipeline for embedded AI.
Python • Rust • PyTorch • NVIDIA CUDA Ecosystem • Docker • Deep Reinforcement Learning • ML Systems
I write about reinforcement learning, ML systems and Edge AI.