Yeronis

Assefa

Hubena

Developed a reinforcement learning agent that learns to play ping-pong against a human player. The AI improves its gameplay by training through repeated interactions, optimizing its paddle movements to successfully return the ball. Implemented reward-based learning, enabling the agent to adapt and enhance its performance over time.

Built a deep neural network model trained on the 'Places' dataset for scene classification tasks and generative adversarial networks (GANs) for image synthesis. The model accurately classifies diverse environments and generates realistic scenes through adversarial training.

Intelligent resume optimization system that analyzes job descriptions and suggests targeted improvements to maximize ATS compatibility and recruiter appeal. Uses advanced NLP techniques to extract key requirements, optimize keyword density, and provide actionable recommendations for content enhancement and formatting improvements.

Advanced computer vision application trained on custom image datasets to recognize and classify objects with exceptional accuracy. Implements state-of-the-art CNN architectures with transfer learning, data augmentation, and real-time inference capabilities for mobile and web deployment.

Sophisticated conversational AI system built on fine-tuned large language models, specialized for domain-specific customer queries and responses. Features context-aware dialogue management, sentiment analysis, and seamless integration with existing customer service workflows for enhanced user experience.

Automated proposal generation system that creates personalized freelance proposals based on client briefs and project requirements. Leverages natural language processing to analyze project descriptions, match relevant skills, and generate compelling, tailored proposals that increase win rates and save time.

Advanced audio analysis system that automatically tags music with moods and emotions using deep learning techniques. Processes audio features, spectrograms, and temporal patterns to classify emotional content, enabling intelligent music recommendation and playlist generation based on user preferences and context.

Advanced real-time motion tracking system using Lucas-Kanade and Farneback optical flow algorithms for precise object movement analysis.

Intelligent panorama creation using SIFT/ORB feature detection, RANSAC-based homography estimation, and seamless image blending.

Comprehensive image processing suite with advanced color space transformations, histogram analysis, and enhancement algorithms.

Deep learning-based semantic segmentation using U-Net and DeepLab architectures for precise pixel-level classification.

Advanced feature detection and matching system for robust panorama generation with automatic keypoint extraction and descriptor matching.