Open Research Projects

If you are a Final Graduation Project student (undergraduate or graduate), you can apply for supervision in one of the research topics listed below. Students will serve as the main researchers, working under the supervision and guidance of a DeepARC research team.

False Assurance in LLM-Assisted Secure Code Generation and Repair

Project Overview

• This project investigates the reliability of Large Language Models (LLMs) when generating, analyzing, and repairing vulnerable source code.

• The research focuses on a critical question: Can we trust an LLM when it claims that generated or repaired code is secure?

• Students will evaluate LLM-generated security assessments and code repairs using real-world vulnerability patterns and automated security analysis tools. Particular attention will be given to cases of false assurance, where an LLM claims that a vulnerability has been identified or fixed, while the underlying security issue remains present.

Research Objectives

1. Evaluate the security reliability of LLM-assisted code generation and vulnerability repair.

2. Characterize and measure false assurance in LLM security assessments and code repairs.

3. Develop an evidence-based benchmark or evaluation protocol for assessing the trustworthiness of LLM-generated security claims.

What You’ll Do

• Generate and analyze code using selected Large Language Models.

• Introduce or identify known software vulnerabilities in source code.

• Ask LLMs to detect, explain, and repair security vulnerabilities.

• Verify proposed repairs using static analysis, vulnerability scanners, testing, or other security validation techniques.

• Compare the LLM's security claims against objective evidence.

• Analyze failure cases where vulnerabilities remain after an LLM reports them as fixed.

• Develop metrics and an experimental framework for measuring security reliability and false assurance.

Technical Requirements

• Basic knowledge of software security and common vulnerabilities.

• Basic knowledge of Python, Machine Learning and Large Language Models.

Expected Results

• A reproducible experimental framework for evaluating LLM-assisted secure coding.

• Quantitative measurements of vulnerability detection, repair effectiveness, and false assurance.

• A characterization of common failure modes in LLM-generated security assessments and repairs.

• An evidence-based benchmark or evaluation protocol for assessing the trustworthiness of LLM security claims.

• Research results targeted for submission to a SCOPUS/WoS-indexed conference or journal.

The Team

• <Open position>

• Karim Elish (Florida Polytechnic University) — Co-advisor

• Manuel Eugenio Morocho-Cayamcela (Yachay Tech University) — Advisor

Source Code

• Source code, experimental configurations, and evaluation artifacts will be made available on GitHub upon completion of the project.

AI for Galápagos: Data-Driven Solutions for Conservation and Sustainability

Project Overview
We invite students to join a collaborative research project between Yachay Tech and the Universidad Internacional del Ecuador (UIDE) focused on applying Artificial Intelligence and Data Science to real-world challenges in the Galápagos Islands.

Students will identify an environmental, ecological, or sustainability-related problem affecting the islands and develop a data-driven solution using relevant datasets and computational techniques.

Objective
The goal is to use AI, Machine Learning, Data Science, GIS, Computer Vision, or other computational approaches to better understand, monitor, predict, or address challenges related to the conservation and sustainable management of the Galápagos.

Students are free to explore the datasets and methodologies most appropriate for their research question.

Structure

• Research Project: Identify a problem, collect relevant data, develop a solution, and evaluate the results.

• Progress Meetings: Regular meetings with faculty from Yachay Tech and UIDE for feedback and guidance.

• Inter-Institutional Collaboration: Students from both universities will collaborate and share ideas and methodologies.

Technical Requirements:

• Intermediate Python.

• Basic knowledge of Artificial Intelligence, Machine Learning or Data Science.

• Interest in applying technology to environmental challenges.

The team:

• <Open position>

• Paulina Vizcaino (Universidad Internacional del Ecuador - UIDE) - Co-advisor

• Iván Reyes (Universidad Internacional del Ecuador - UIDE) - Co-advisor

• Manuel Eugenio Morocho-Cayamcela (Yachay Tech University) - Advisor

Expected results:

• Research results may include AI models, data analysis, environmental monitoring tools, predictive models, geospatial analyses, or other data-driven solutions.

• Results may be submitted to SCOPUS/WoS-indexed journals and conferences.

Source code:

• Avaliable on GitHub upon project completion.

Data Science and Business Intelligence in the Ecuadorian Stock Market

Description:

• This project aims to analyze data from the Ecuadorian Stock Exchange (Bolsa de Valores del Ecuador) and develop time-series forecasting models to identify patterns, trends, and potential signals in the Ecuadorian stock market.

• The project will explore how Data Science and Business Intelligence techniques can be applied to financial market data to support data-driven investment analysis and decision-making.

Technical Requirements:

• Basic knowledge of time-series forecasting.

• Intermediate-level in Python programming.

The team:

• <Open position>

• Andrés Navas (Universidad de las Américas - UDLA) - Co-advisor

• Manuel Eugenio Morocho-Cayamcela (Yachay Tech University) - Advisor

Expected results:

• Development and evaluation of time-series forecasting models (probabilistic, machine learning, deep learning, transformers-based, etc.).

• Data-driven analysis of trends and patterns in the Ecuadorian stock market.

• Results targeted for submission to a SCOPUS-indexed conference or journal.

Source code:

• Source code will be made available on GitHub upon completion of the project.

Machine Learning as an Enabler of Next-Generation Wi-Fi networks

Technical requirements:

• This project explores the application of Machine Learning techniques to improve the performance, efficiency, and intelligence of next-generation Wi-Fi networks.

• Students will investigate how machine learning can be used to address challenges in wireless networks, such as network optimization, resource allocation, traffic prediction, quality-of-service management, or anomaly detection.

Technical Requirements:

• Prior knowledge of wireless networks.

• Basic knowledge of Machine Learning.

• Intermediate-level Python programming.

The team:

• <Main Researcher>

• Juan Pablo Astudillo León (Yachay Tech University) - Co-advisor

• Manuel Eugenio Morocho-Cayamcela (Yachay Tech University) - Advisor

Expected results:

• Development and evaluation of Machine Learning models for wireless networking.

• Experimental results demonstrating potential improvements in network performance or efficiency.

• Results to be submitted to a scientific conference or journal.

Source code:

• Source code will be made available on GitHub upon completion of the project.

Transport Obstruction in Diffusion Tensor Imaging for AI

Technical requirements:

• Python (basic voxel data handling), PyTorch

What you’ll do:

• Implement methods from a research paper (DTI analysis)

• Process medical imaging data (brain scans)

• Compute and visualize advanced features (e.g., spatial coherence metrics)

• Test these features in ML/DL models for disease detection

Potential applications:

• Early detection of neurological disorders

• Improved medical image analysis using AI

Available data:

• High angular resolution diffusion MRI

• Kick-off paper

• Derived features (e.g., FA, custom metrics from paper)

The team:

• <Open position>

• Manuel Eugenio Morocho-Cayamcela (Yachay Tech University) - Advisor

• Daniel Alayón-Solarz (ServiceNow)- Co-advisor

Expected results:

• Implementation and validation of the DTI analysis methods described in the reference research paper.

• Extraction and visualization of diffusion MRI features, including FA and spatial coherence metrics.

• Evaluation of these features using Machine Learning and Deep Learning models for neurological disease detection.

• Comparative analysis of the predictive value of conventional and proposed imaging features.

• Results targeted for submission to a scientific conference or journal.

Source code:

• Source code will be made available on GitHub upon completion of the project.