Thomas Donald Jarvis PRO

AI & ML Interests • Behavioral AI systems • Local-first AI architectures • Vision-language models (VLMs) • Compact / efficient transformer architectures • Sparse Mixture-of-Experts (MoE) systems • Calibration-aware machine learning • Confidence-governed AI decision systems • Multimodal reasoning systems • Financial / trading AI systems • Volume Price Analysis (VPA) AI research • AI alignment through deterministic governance layers • Edge AI deployment and optimization • LoRA / QLoRA fine-tuning • Behavioral compression and “crystal”-based learning • Corrective training loops and targeted refinement • Model merging and adapter tournaments • Open-source AI research • Autonomous AI systems engineering • Quantitative finance + ML integration • Lightweight inference optimization • FP8 / low-precision inference systems • Synthetic dataset generation • AI calibration and uncertainty estimation • Specialized decision intelligence systems • Scratch-trained transformer research • AI routing and expert specialization • Human-AI systems architecture • Efficient multimodal inference on consumer hardware