KAP THANG
AI/ML Research Scientist & Full-stack Developer
Summary
A motivated Canadian AI/ML researcher and full-stack developer completing an M.Sc. in Computer Science (AI Stream) at the University of Windsor in April 2026. My thesis focuses on LLM-based synthetic review generation and NLP analysis, and my research includes developing neural models with PyTorch—GNNs, Transformers, and Bayesian networks. My soft skills include decision-making, creativity, attention to detail, emotional intelligence, problem-solving, teamwork and communication. I've deployed 3 React apps with more to come, including AI-powered apps. Seeking a full-time position as an ML Engineer, AI/ML Research Scientist, Data Scientist, Software Engineer, or related role starting May 2026. Fully authorized to work in Canada and US (via TN visa).
Skills
Languages
Python • TypeScript • JavaScript • Kotlin • Java • C/C++ • HTML • CSS • SQL • NoSQL • Bash
Technologies
Git • GitHub • Docker • Expo • Node.js • GenAI/LLM • MCP • REST API • Azure • GCP • CI/CD
Frameworks & Libraries
PyTorch • Scikit-learn • Pandas • NumPy • React • React Native • Next.js • Express.js • Nest.js • Tailwind • FastAPI • Hugging Face • LangChain
Languages
Portfolio
kapthang.devInterests
Reading about the future of AI & robotic news, researching in AI/ML topics, building cross-platform web and mobile apps using React, investing in the future, watching movies, and playing games
Experience
AI/ML Research Assistant
Fani's Lab, University of Windsor
- Developed ML models using PyTorch-based framework OpenNMT, achieving 82x improved performance over baselines across 4 large-scale datasets, contributed and collaborated in an open-source library (OpeNTF) for a directed ML course
- Optimized and implemented NMT seq-to-seq models through fine-tuning and explored to identify best-performing architectures for the benchmark library
- Integrated implicit dataset support into the LADy library, extracted annotations and enhanced NLP analysis, with improved synthetic reviews achieving 93% performance parity vs. real datasets
Teaching & Graduate Assistant
SCS, University of Windsor
- Communicated with 250+ students and delivered technical instruction and code reviews for web development (JavaScript, React, PHP) and systems programming (C, Linux, Bash) courses
- Analyzed and assisted with debugging and feedback on course assignments, managed and led 30+ lab sessions, supervised and inspired students/mentees
CartDash Founder & Full-stack Developer
kapthang.dev
- Building a 30+ features cross-platform shopping list app using Expo and React Native (iOS, Android, Web), including zero-knowledge encryption, utilizing TypeScript and Convex backend for offline-first architecture, with CRDT-based real-time sync for collaborative cart sharing
Education
M.Sc. in Computer Science (AI Stream)
University of Windsor
- Thesis: Context-Engineered Reviews Architecture (CERA) for LLM-based synthetic review generation and NLP analysis
- Coursework: Deep Learning, Recommender Systems, Graph Neural Networks
Honours B.Sc. in Computer Science (Software Engineering)
University of Windsor
- Coursework: Data Structures & Algorithms, Database Systems, Software Architecture, Advanced Website Designs, Systems Programming
Publications
Translative Neural Team Recommendation: From Multilabel Classification to Sequence Prediction
ACM SIGIR '25
Kap Thang, Hawre Hosseini, Hossein Fani
Proposed a seq-to-seq approach using transformers for neural team recommendation, reformulating the problem from multilabel classification to sequence prediction and demonstrating superior performance across four large-scale datasets.