Joseph Siet Jiun Wen

An enthusiastic, independent forefront practitioner, teacher and learner-on-the-way-to-expert individual, hinting mainly in the fields cutting edge AI development (i.e. the proprietary models undertaken by OpenAI, Anthropic, Google DeepMind, Perplexity AI, and the open-sources including DeepSeek, Qwen and Llama's dedicated model architectures and/ or tokenization strategies) and the underlying historical trajectories of germinations in LLMs (Transformers, Mixture-of-Experts, Supervised Fine-Tuning, Reinforcement Learning (w/ Human Feedback, aka RLHF), Chain-of Thoughts Reasonings, Simulated Reasonings). My specializations fall within the contexts of Artificial Intelligence, peculiarly deep learning, machine learning and mathematical modelling for Computer Vision. My personalized brand stems from a strong interest and relentless pursuits of research piques in terms of applying the fundamental Math. and Sci. models (Percolation Theories, 2ND Law of Thermodynamics, Abstract Algebras w/ Advanced Calculus, Topologies, Tensor-Approximations, Spectral-Analysis of Random Matrix Theory) into the realm of Machine Learning and Deep Learning for Computer Visions. In the lexicon of deep learning and machine learning, I spent 3-4 years into ruminating the fundamentals underpinning the deep and shallow Computer Vision Models, inclusive of various architectures like ViT and MoE w/ their use cases for practical adoptions. I completed a Final Year-end project and Master Project, which contributes to **Seven (7) Publications in the Biomedical, Math. and Sci. / Comp. Sci. Journals and Conferences** till the period before August 2025. (My FyP hints on applying the opensource (Python, OpenCV-Python, TensorFlow) deep learning modules and image processing framework into the research works for Automated Optical Inspection (AOI) prior to underscore the potential defects in PCBs, whereas the Master Project aims for producing the premier antecedences in biomedical image engineering solutions prior to demarcate the grading nuances for cancer terminality and malignancies of breast cancer patients w/ the helps of MATLAB licensed CV software features.)