about

Ruize Ma builds AI systems that can be inspected.

This portfolio presents projects as product-like systems rather than static screenshots. The goal is to make model behavior legible: what goes in, what happens in the middle, what comes out, and where the limits are.

This version uses a Next.js frontend with a FastAPI backend wrapper, so the interface can preserve the original project logic while making the system traces easier to inspect.

Focus

AI

agent workflows and model reasoning

Interface

UX

dashboards for readable systems

Delivery

fullstack

Next.js frontend + FastAPI backend

working style

Product clarity for technical systems

planner

LLM Parser

query JSON

Converts the natural-language roster request into team, goal, top-k, recent-game window, and availability filters.

planner

GPT5Model.plan

plan JSON

Analyzes the stem and choices, extracts keywords, selects facts needed, and proposes biomedical tools.

classifier

Block 2 Router

segment

Uses known smoker status directly, or estimates smoker probability for unknown status.

stack

Next.js App RouterTypeScriptTailwind CSSFramer Motionlucide-react