UX Research × AI Product Prototyping × Consumer Hardware

I am a UX researcher working in consumer electronics, and an independent builder exploring long-term human–AI interaction, local-first AI systems, and tools that solve problems I actually encounter.

My work tends to follow the same path:

observe a real problem → build something to test it → find the deeper question → iterate

I am not trying to present myself as a software engineer. GitHub is where I keep the things I build to understand products, AI systems, and my own workflows more concretely.

An early local-first personal-AI engineering prototype combining local LLM inference, an always-on agent runtime, long-term memory, custom plugins, conversation processing, and fine-tuning experiments.

Building it led to a broader research question: if personal AI is meant to learn from someone over time, should interaction history really be reduced to independent Q&A pairs? The public repository is now maintained as the engineering origin and project lineage; later scientific work is maintained separately during the research/submission process.

A privacy-conscious, local-first document ingestion and retrieval pipeline for turning personal documents into structured Obsidian Markdown and searchable Qdrant knowledge.

The public version deliberately separates reusable code from private runtime data, with synthetic tests, release guards, contributor boundaries, and a public roadmap.

A small defensive tool for checking NVIDIA API lifecycle and compatibility assumptions before automation depends on them. Built around conservative validation rather than optimistic version guessing.

Before the larger AI systems, I was already building small tools for my own needs and curiosity:

• Fractal Scripts — Mandelbrot, Julia, Koch snowflake, Sierpiński triangle and fractal-tree experiments
• GPT HTML Dialogue Extractor — extract large exported ChatGPT conversations into separate text files
• EXIF Batch Rename — organize photo libraries by capture time
• File Organizer — duplicate-file and empty-folder cleanup utilities

My independent work on personal AI gradually moved from system building into questions about longitudinal interaction data, thought trajectories, and how AI systems learn when to use what they have learned.

• What if your AI could grow with you?
• Personal AI isn't about answers — it's about thought trajectories

Current empirical HCI research is kept separate from this public portfolio while manuscripts and reproducibility packages are under active research/submission workflows.

My professional background is in consumer-electronics UX research and market research. I work across qualitative and quantitative methods, large-scale behavioral / after-sales data, user segmentation, product experience diagnosis, and upstream decision support.

That background is also why I build: prototypes give me another way to test assumptions about what a product or AI system can actually do, where it breaks, and what problem is worth solving next.