About Me
I am a Machine Learning and AI Engineer with a strong engineering foundation and a Master’s degree in Computer Science (Artificial Intelligence). My background combines hands-on experience in advanced process engineering with end-to-end development of production-ready AI and data systems. I specialize in transforming real-world, complex systems into scalable, data-driven solutions.
My work spans the full lifecycle of modern machine learning applications, including data engineering, feature extraction, model development, deployment, and monitoring. I have designed and deployed multiple full-stack AI solutions, including an LLM-powered healthcare question-answering system utilizing Retrieval-Augmented Generation (RAG) over large-scale datasets, supported by automated MLOps pipelines that enhance reliability and deployment efficiency. I also have experience building applied machine learning models for classification, prediction, and anomaly detection, using techniques such as XGBoost, Random Forest, deep learning, and natural language processing. My projects emphasize measurable impact—improving accuracy, reducing manual effort, and delivering insights through intuitive, data-driven interfaces.
I am passionate about building scalable, reliable, and production-ready AI systems using modern tools, such as Python, PyTorch, TensorFlow, Docker, Kubernetes, and MLflow, as well as cloud platforms including AWS, GCP, and Azure. I enjoy solving complex problems and turning data into practical, impactful solutions.
Education
- Master of Science in Computer Science (AI Concentration) – Troy University
- BEng in Electronic & Electrical Engineering – University of Sunderland
Technical Skills
- Programming Languages: Python, C/C++, JavaScript, SQL, HTML
- Machine Learning & AI: Deep Learning, Neural Networks, NLP, LLMs, Decision Trees, Random Forest, XGBoost, SVM, Gradient Boosting
- Frameworks & Libraries: TensorFlow, PyTorch, Keras, Scikit-learn, Pandas, NumPy, Flask, D3.js, LangChain, Hugging Face, RAG
- MLOps & Cloud: AWS SageMaker, GCP Vertex AI, Azure ML, Docker, Kubernetes, CI/CD Pipelines
- Data & Visualization: Jupyter Notebook, Tableau, Power BI, Hadoop
Work Experience
- Evaluated and validated LLM and NLP systems to strengthen semantic understanding, intent alignment, and overall model reliability across diverse real‑world inputs.
- Conducted structured error analysis to identify model failure patterns, bias, and edge cases, reducing recurring model errors by 20%.
- Supported supervised and reinforcement learning workflows by producing high‑quality, guideline‑compliant datasets used for model training and iterative refinement.
- Monitored post‑deployment model behavior to detect drift, anomalous outputs, and degradation trends early, contributing to more stable and predictable system performance.
- Applied human‑in‑the‑loop (HITL) evaluation methods to refine model reasoning, reduce inconsistencies, and enhance safety and compliance across model outputs.
- Analyzed user intent, contextual meaning, and semantic relevance to improve ranking signals, retrieval quality, and downstream AI‑driven decision processes.
- Interpreted complex evaluation frameworks and applied them consistently to ambiguous data, ensuring high‑precision judgments that support model alignment and quality assurance.
- Led data-driven process optimization for high-volume manufacturing lines, improving yield, cycle time, and defect rates through statistical analysis.
- Built structured datasets from sensor data, equipment logs, and quality metrics to support root cause analysis and trend detection.
- Applied regression analysis, DOE, and multivariate techniques to identify key process drivers and validate improvements.
- Developed Python-based tools and dashboards to automate KPI tracking and engineering analysis.
- Supported automation and equipment validation by evaluating process capability (Cp, Cpk) and system repeatability.
- Collaborated with manufacturing, quality, R&D, and supply chain teams to support NPI and scalable production ramp-ups.
- Authored SOPs, control plans, PFMEA, and validation reports to ensure repeatable and compliant manufacturing execution.
TELUS Digital | AI Engineer
Techtronic Industries (TTI) | Advance Process Engineering (APE) Engineer
Projects
- Healthcare Chatbot
- Threat Detection using Machine Learning
- Global Population Prediction
- Sleep Quality Prediction
- Smart Trash System
- Anti-Burglar System House Connected to Smartphone
A chatbot is being designed and implemented using Large Language Models (LLMs) and Retrieval-Augmented Generation (RAG) from 250,000+ data records to provide accurate and context-aware responses. The system features an end-to-end MLOps pipeline for data ingestion, preprocessing, model training, API serving, and continuous deployment, utilizing Docker and CI/CD workflows. FastAPI supports scalable RESTful inference, while Streamlit provides an interactive web interface. To enhance retrieval precision and minimize hallucinations, vector search (FAISS) and prompt engineering techniques are employed.
Technologies Used: Python, Flask, LangChain, FAISS, PyTorch, FastAPI, Docker, CI/CD
Developed an NLP-powered machine learning application to identify harmful online content in real time. Using a dataset of 72,000+ social media posts, I applied advanced text preprocessing (tokenization, stemming, TF-IDF) and engineered semantic, syntactic, and sentiment features. Multiple models were tested, with Logistic Regression, Random Forest, and XGBoost achieving the highest performance at 92% accuracy. The system was deployed as a Flask web app, providing live classification, offensive word highlighting, and a bullying severity score for interpretability.
Technologies Used: Python, Scikit-learn, XGBoost, Flask, Pandas, NumPy
Created a forecasting tool to analyze and predict demographic trends from 1960 to 2023 with 95% prediction accuracy. The dashboard featured interactive maps, line charts, and bar charts, enabling intuitive exploration at both global and country levels. Built using Python, D3.js, Tailwind CSS, and Pandas, the system combines statistical forecasting with rich visualizations, making demographic data accessible to policymakers, researchers, and the public.
Technologies Used: Python, Pandas, NumPy, D3.js, ETL Pipelines, Data Visualization, Flask
Designed an AI system to predict sleep quality using 40,000+ records of lifestyle and biometric data, including sleep duration, stress, heart rate, and daily activity. Experimented with Logistic Regression, Random Forest, and deep learning architectures (CNNs and RNNs), achieving 96% accuracy. Delivered as a Flask-based dashboard, the tool allows users to input lifestyle data and receive predictions along with personalized health recommendations, demonstrating the power of machine learning for wellness applications.
Technologies Used: Python, Scikit-learn, TensorFlow, CNNs, RNNs, Flask
Created a solar-powered smart trash bin with automatic lid control and real-time waste monitoring using ultrasonic sensors. Designed 3D mechanical models in SolidWorks, manufacturable layouts in AutoCAD, and control circuitry via PCB tools and Multisim. Integrated renewable energy for autonomous outdoor use, demonstrating innovation in sustainable IoT design.
Technologies Used: Arduino, SolidWorks, AutoCAD, PCB Design, Multisim, Solar Power, Integration
Developed a home security system using sensors and Arduino, connected to smartphones for remote monitoring and control. Designed wiring diagrams, house models, and PCB layouts to ensure efficient installation and operation. Enhanced system functionality through iterative testing, making it a practical and affordable IoT security solution.
Technologies Used: Arduino, PCB Design, AutoCAD, Multisim, Embedded Systems
Certifications
Languages
- Language: English and Vietnamese