Mohamed Amine Oulahyane

Associate Degree in Computer Science

Feb 2023 – Present

Citrus College, Azusa, CA

Currently pursuing an Associate Degree in Computer Science with a strong focus on software development and artificial intelligence. Expected transfer date: Fall 2025.

FinAid: Personal Financial Assistant - AI-Powered Financial Management Platform

2024

Python, Next.js, Django, & GPT Technology

• Financial Data Analysis: Leveraged AI to analyze and categorize bank statements, providing users with real-time insights into their spending habits.
• Natural Language Processing (NLP): Integrated advanced NLP models to enable users to query their financial data and receive personalized advice.
• Responsive Web Design: Developed a user-friendly interface using Next.js, ensuring seamless experience across devices.
• AI-Driven Insights: Employed LLMs and GPT to generate actionable financial insights, helping users optimize their budgeting and saving strategies.
• Web Development: Engineered both frontend and backend systems with Next.js and Django, enhancing the platform's performance and scalability.
• User Experience Enhancement: Conducted extensive testing and user feedback sessions to refine the platform’s UX/UI, improving user engagement and satisfaction.

Pac-Man AI Video Game - Expertise in Deep Learning and Reinforcement Learning

Jan 2024

Python & TensorFlow

• CNNs & LSTMs: Used CNNs for visual processing and LSTMs for strategic decision-making, enhancing AI efficiency.
• A3C Algorithm: Implemented A3C in the Gymnasium environment for robust and parallel training of AI agents.
• Unity & TensorFlow Integration: Combined Unity for game development with TensorFlow for neural network operations.
• Performance Optimization: Conducted extensive training and hyperparameter tuning for optimal AI performance.
• AI Visualization: Created real-time visualizations of AI decisions to showcase strategy and engagement.

Autonomous Car Racing Simulation - Real-Time Stock Monitoring & Analysis

Jul 2024

Python & TensorFlow

• Convolutional Neural Networks (CNNs): Employed CNNs for real-time image processing and path recognition.
• Proximal Policy Optimization (PPO): Utilized PPO for training the AI to navigate complex racing tracks efficiently.
• Real-World Simulation: Conducted extensive simulations to test AI performance in various racing scenarios and optimize driving strategies.
• Team Collaboration: Worked with a team to refine the AI's decision-making capabilities and enhance adaptability to dynamic environments.
• Performance Optimization: Engaged in iterative testing and hyperparameter tuning to achieve optimal AI performance.

Mountain Car Simulation - Reinforcement Learning for Continuous Control

2024

Python & Gymnasium

• Deep Q-Network (DQN): Implemented DQN to teach the AI how to navigate the Mountain Car environment by overcoming gravity-based challenges with optimal policy exploration.
• Hyperparameter Tuning: Adjusted learning rate and discount factor to achieve stable learning and efficient performance.
• Exploration vs. Exploitation: Experimented with different ε-greedy strategies to balance the AI’s exploration of new actions and exploitation of known rewards.
• Performance Metrics: Tracked total reward per episode and monitored the car's performance over time to ensure training improvements and efficient learning.

Blackjack AI - Reinforcement Learning in a Card-Based Environment

2024

Python & Gymnasium

• Q-Learning Algorithm: Trained an AI agent using Q-learning to make optimal decisions in a simplified blackjack game environment.
• Reward Maximization: Focused on maximizing the cumulative reward by optimizing decisions based on the current state and future expectations.
• State Space Exploration: Explored various states of the game (cards dealt, dealer's cards) to enable the AI to make accurate, data-driven decisions.
• Policy Improvement: Continuously improved the policy through iterative updates, refining the AI’s decision-making ability with each game episode.