RL-GPT
Prototype / ResearchReinforcement learning experiments for GPT-style tuning with custom reward functions.
Work
A mix of production-focused engineering, research prototypes, and AI experimentation across LLM systems, computer vision, and simulation design.
Reinforcement learning experiments for GPT-style tuning with custom reward functions.
Wiki-scraper and JSON builder for context-rich lorebook generation.
Visual evolution experiments using computer vision, VQGAN, and CLIP.
NOAA buoy-cam harvester using computer vision and geospatial filtering to capture ideal sunset footage.
CLIP-powered scene grouping for video and image organization using DBSCAN and HDBSCAN.
Gesture-first operating shell using pose estimation and fuzzy logic concepts.
Prompt-engineering research on token density and controllable LLM output.
Python simulation and game project.