Portfolio

Publications

A Scalable Kernel Approach to Reinforcement Learning

We propose a self-contained, detailed, description of a scalable standardized kernel (RKHS) approach to popular reinforcement learning algorithms, where agents interact with environments having continuous states and discrete actions spaces, dealing with possibly unstructured data. These algorithms, namely Q-Learning, Actor Critic, Q-Value Gradient, Hamilton-Jacobi-Bellman (HJB) and Heuristic Controls, are implemented with a RKHS library using default settings. We show that this approach to reinforcement learning is accurate, robust, sample efficient and versatile, as we benchmark our algorithms in this paper on simple games and use them as a baseline for our applications.

Research paper

Making a fast RL env in C with pufferlib

Medium article

What are GAEs (Generalized Advantage Estimations) in Reinforcement Learning?

Medium article

Policy evaluation in Reinforcement Learning: Understanding Eligibility traces

Medium article

Projects & Contributions

PufferLib

Pufferlib is a high-performance toolkit for research and industry with optimized parallel simulation, environments that run and train at 1M+ steps/second, and tons of quality of life improvements for practitioners. Puffer Website

EverRAG

EverRAG transforms your Evernote into an intelligent, searchable database.

Github repository

Risk global domination RL bots

Making my custom gymnasium environment on the game of RISK global domination, and solving it using RL algorithms

Github repository

Reinforcement Learning: An Introduction. My Answers and solutions to exercises.

Here I wrote down my answers while I studied the book from Barto,Sutton. Programming exercises as well with custom gym environments.

Github repository

Kaggle

Code from my kaggle competitions and dataset analysis.

Github repository

RL_playground

A repository containing different RL algorithms implementations on basic environments

Github repository