hi! i'm namila rahmani

RL DDQN Poker Agent

Implemented double deep Q-Network (DDQN) RL technique with experience replay and an abstracted state and action space to train an agent for 6 handed poker as a team.

View GitHub repo here

FPGA Color Robot

Built robot to navigate towards a specific color object using FPGA, servo motors, particle photon board, and python openCV as a team for CS 429H (Computer Architecture Honors) final project.

View presentation here

View GitHub repo here

Paylocity Internship (Summer 2024)

SWE Intern for document library / fillable forms team - utilized React.js, C# .NET Core, and SQL to implement new expiration features for compliance forms; worked with Agile methodology, Cypress, TeamCity & Octopus to complete the full development to deploy pipeline. Refined APIs to improve latency for document library with > 1M API calls/day; helped mitigate latency spike resulting in performance improvement of 92%. Modified databases and workengine to properly track assignment IDs from integrator RabbitMQ messaging.

Learn more about paylocity here

Operating System

Collaborated with Turing Scholars class to implement fully functional operating system in C supporting synchronization primitives, task scheduling, I/O, file systems, virtual memory, user programs, and system calls, with capability to run the Doom game and networking features. As part of the I/O team, designed and implemented an interface for integration with other teams.

I/O Team Presentation

RL Snake Player

Applied reinforcement learning to the 2D grid-based game called Snake, including training algorithms of Q-learning with linear function approximation, Deep Q-Network (DQN), and Double Deep Q-Network (DDQN) with experience replay to observe how these techniques compared in performance on a partially observable Markov decision process (POMDP) of the traditional version of the game. Final Project for CS 394R (Graduate Reinforcement Learning) with partner.

View github repo here

Tetris

Implemented the common Tetris game, where pieces fall through a 2d grid of squares and full rows are cleared, and created an AI Tetris Brain, which was able to play the game independently through ranking outcomes. Implemented in Java and tested with JUnit