EANational Service Personnel 1 Students Life & Enga
Emmanuel Agyei Brewu portrait

Early-career creator building practical skills

I am launching my professional journey by learning tools and completing portfolio work that demonstrates real ability.

View my workMessage Me

Work samples

Short projects and learning exercises that illustrate the tools and workflows I am practicing as I prepare for professional work.

  • UNDERGRADUATE THESIS, Ashesi University
    UNDERGRADUATE THESIS, Ashesi University

    Evaluating the Computational Overhead and Latency in AI-Based DDoS Detection Systems  Designed, implemented, and evaluated LUCID, a lightweight CNN-based DDoS detection system for computationally constrained environments.  Used Python, TensorFlow/Keras, Kali Linux, multiprocessing, model pruning, quantization, and Optuna-based hyperparameter tuning to develop and optimize the detection system.  Simulated DDoS attacks in virtualized environments and evaluated model performance and computational overhead.  Achieved approximately 90% reduction in training time and 35% improvement in inference speed through model and system-level optimization.  Integrated multithreaded packet capture and real-time prediction with alerting and logging to support practical edge deployment.

    PythonTensorFlow/KerasKali Linuxmultiprocessingmodel pruningquantizationOptuna
  • COMPUTER VISION & DEEP LEARNING PROJECT, Ashesi University
    COMPUTER VISION & DEEP LEARNING PROJECT, Ashesi University

    Group 13 Age Detection 1 Facial Age Detection  Developed a CNN-based facial age-group classification system using the UTKFace dataset.  Pre-processed facial images through grayscale conversion, resizing to 150D7150 pixels, reshaping, and pixel normalization.  Built the model using the Keras Functional API, with convolutional, pooling, dense, and dropout layers for 14 age- group classifications.  Applied GridSearchCV and SciKeras to tune model parameters and evaluated performance using accuracy, loss curves, confusion matrices, and classification reports.  Developed a Streamlit application enabling image upload and webcam-based prediction, using OpenCV and the trained Keras model for inference.

    Keras Functional APIGridSearchCVSciKerasStreamlitOpenCVKeras

About me

I am launching my professional journey by learning tools and completing portfolio work that demonstrates real ability.

I am Emmanuel Agyei Brewu. I am at the start of my career and building skills and projects to launch into professional work. I focus on learning practical tools and completing portfolio work that shows what I can do.

Experience

National Service Personnel 1 Students Life & Engagement

ASHESI UNIVERSITY, Berekuso · August 2025 1 August 2026

Volunteer Mathematics Teacher

TWAFOMAN SENIOR HIGH SCHOOL, Berekuso · September 2024 1 April 2026

Education

Ashesi University College Berekuso, Eastern Region

BSc. Computer Science · Class of 2025

Skills

Early-career, building practical skillsPortfolio projects and hands-on learningOpen to entry-level opportunities
Back to work

UNDERGRADUATE THESIS, Ashesi University

Evaluating the Computational Overhead and Latency in AI-Based DDoS Detection Systems  Designed, implemented, and evaluated LUCID, a lightweight CNN-based DDoS detection system for computationally constrained environments.  Used Python, TensorFlow/Keras, Kali Linux, multiprocessing, model pruning, quantization, and Optuna-based hyperparameter tuning to develop and optimize the detection system.  Simulated DDoS attacks in virtualized environments and evaluated model performance and computational overhead.  Achieved approximately 90% reduction in training time and 35% improvement in inference speed through model and system-level optimization.  Integrated multithreaded packet capture and real-time prediction with alerting and logging to support practical edge deployment.

PythonTensorFlow/KerasKali Linuxmultiprocessingmodel pruningquantizationOptuna
View all work
Back to work

COMPUTER VISION & DEEP LEARNING PROJECT, Ashesi University

Group 13 Age Detection 1 Facial Age Detection  Developed a CNN-based facial age-group classification system using the UTKFace dataset.  Pre-processed facial images through grayscale conversion, resizing to 150D7150 pixels, reshaping, and pixel normalization.  Built the model using the Keras Functional API, with convolutional, pooling, dense, and dropout layers for 14 age- group classifications.  Applied GridSearchCV and SciKeras to tune model parameters and evaluated performance using accuracy, loss curves, confusion matrices, and classification reports.  Developed a Streamlit application enabling image upload and webcam-based prediction, using OpenCV and the trained Keras model for inference.

Keras Functional APIGridSearchCVSciKerasStreamlitOpenCVKeras
View all work
Back to work

STATISTICS RESEARCH PROJECT, Ashesi University

The Impact of Corruption on Governance in Ghana  Analyzed Round 8 Afrobarometer Ghana survey data using R and RStudio, investigating relationships between perceived corruption, democracy, and trust in the police.  Defined research questions, variables, and hypotheses, and structured categorical survey data for statistical analysis.  Created contingency tables and data visualizations to examine relationships among the study variables.  Conducted chi-square tests of independence and post-hoc analysis, interpreting statistical significance and communicating findings.  Evaluated methodological limitations, including the inability of observational data to establish causality, and identified directions for further research.

RRStudio
View all work
Back to work

INFORMATION & SYSTEMS SECURITY PROJECT, Ashesi University

Worked in a four-member team to design and implement a secure relational database system.  Applied database security principles including role-based access control, authentication, SQL security, and controlled access to data.  Contributed to system design, implementation, testing, and documentation.

View all work
Back to work

Campus News Portal 1 Full-Stack Web Application

Developed and publicly deployed a full-stack platform for publishing and managing student-written news using HTML, CSS, JavaScript, PHP, and MySQL. Designed and normalized relational tables for administrators, categories, and posts, supporting structured content management. Implemented authentication, review workflows, and role-based publishing to control how content moved through the platform. Integrated frontend elements with PHP/MySQL backend services to support dynamic content and administrative workflows.

HTMLCSSJavaScriptPHPMySQL
View all work
Back to work

Database Security Design 1 QuickShop E-Commerce Platform

Collaborated in a four-member team to design a secure relational database covering users, products, orders, and order details. Implemented bcrypt password hashing, secure session tokens, role-based access control, SQL GRANT/REVOKE privileges, and least-privilege access. Used parameterized queries and stored procedures to reduce SQL injection risks and improve the reliability of database interactions.

bcryptparameterized queriesstored proceduresSQL
View all work
Back to work

LUCID 1 Lightweight AI-Based DDoS Detection System

Designed and implemented a lightweight CNN-based detection system focused on computational efficiency and real-time responsiveness. Integrated live multithreaded packet capture, prediction, and logged alerts for real-time monitoring. Applied pruning, quantization, multiprocessing, and Optuna-based hyperparameter tuning, reducing training time by 90% and improving inference speed by 35%.

CNNpruningquantizationmultiprocessingOptuna
View all work
Back to work

IT HUB / HURUPAY 1 Software Solutions

Coordinated cross-functional teams developing two software solutions, managing project scope, task delegation, resources, and presentations. Worked collaboratively with team members to develop solutions while maintaining alignment with academic and industry expectations. Delivered projects 15% under budget and ahead of schedule.

View all work