Bryium Onyancha

I am a research-focused Master's student in Computer Science with hands-on experience in machine learning, including transformer models, reinforcement learning, recursive language knowledge, and scalable GPU-accelerated ML pipelines.

Alongside my research work, I am a software engineer skilled in building modern web applications, robust backend APIs, and mobile applications using contemporary frameworks and tools.

My research focus areas include Large Language Models, Diffusion Models, Multimodal Systems, Recursive Language Knowledge, and Computer Vision — with a long-term goal of developing scalable, compute-efficient architectures for next-generation LLMs and multimodal agents.

I conduct research on efficient training, adaptive inference, and reinforcement-learning–based reasoning in modern foundation models. I also focus on post-training techniques such as RLHF, RLAIF, and iterative refinement for alignment, controllability, and deep reasoning. Additionally, I explore model compression, architecture search, and training-time optimizations to reduce compute while preserving frontier-level performance.

My methods are applied to computer vision tasks like object recognition, image generation, and scene understanding, as well as predictive modeling problems in structured or sequential data.