WGCV

Projects

Where imagination unleashes innovation: A showcase of ideas

Tab Recall Simplified

Tab Recall Simplified

Project distills large LLMs into compact models for efficient web page summarization. Leverages transfer learning with T5-small (60.5M params), Pegasus-xsum (570M), and BART-large (406M). Custom dataset from diverse sources (nytimes.com, GitHub) used for fine-tuning.

Learn more →
Social Media Analysis for the Mayor of Guayaquil 2019 with LSTM models

Social Media Analysis for the Mayor of Guayaquil 2019 with LSTM models

Using NLP and deep learning to analyze 11,230 tweets from Guayaquil's 2019 mayoral candidates. Implemented an LSTM model for sentiment classification, achieving 85% accuracy in political tweets analysis.

Learn more →
From Spanish Audio to Insight: Text Transcription and Sentiment Analysis

From Spanish Audio to Insight: Text Transcription and Sentiment Analysis

Utilized Facebook's wav2vec2-large-xlsr-53 for Spanish audio-to-text transcription and the Robertuito model for sentiment classification (positive, neutral, negative), integrating Transformers and Gradio for a comprehensive solution.

Learn more →
Custom Firmware for the Artillery 3D Printer

Custom Firmware for the Artillery 3D Printer

This C-language firmware for Artillery Sidewinder printers upgrades functionality with M600 filament change, bed leveling, Gcode console, PID temperature control, and more, adding premium features to a mid-tier 3D printer without hardware modifications.

Learn more →
AI-Powered Whiteboard: Predicts handwritten letters and converts to text with TensorFlow.js

AI-Powered Whiteboard: Predicts handwritten letters and converts to text with TensorFlow.js

An AI-powered whiteboard using TensorFlow.js, Canvas, HTML, and JavaScript. Real-time CNN model for accurate handwritten letter classification and autocorrection.

Learn more →