Skip to content
View Ryuichi-Yamafuji-Lun's full-sized avatar
🎯
Focusing
🎯
Focusing
6 followers · 19 following

Achievements

Achievement: QuickdrawAchievement: Pair ExtraordinaireAchievement: YOLOAchievement: Pull Shark

Achievements

Achievement: QuickdrawAchievement: Pair ExtraordinaireAchievement: YOLOAchievement: Pull Shark

Block or report Ryuichi-Yamafuji-Lun

Block user

Prevent this user from interacting with your repositories and sending you notifications. Learn more about blocking users.

You must be logged in to block users.

Maximum 250 characters. Please don't include any personal information such as legal names or email addresses. Markdown supported. This note will be visible to only you.
Report abuse

Contact GitHub support about this user’s behavior. Learn more about reporting abuse.

Report abuse
Ryuichi-Yamafuji-Lun/README.md

Typing Intro

Software Engineer – Backend • AI • Systems

M.S. Computer Science @ USC

Bridging the gap between AI models and high-performance infrastructure.

🌐 Connect

LinkedIn   Portfolio   Google Scholar   Email

🛠 Tech Stack


👤 About Me

  • 🔭 Core Focus: Specializing in Scalable Backend Systems, AI Agents, and HPC.
  • ⚙️ Systems Engineering: Optimized concurrency protocols in C++, achieving 610K TPS via elastic locking mechanisms.
  • 🧠 AI & ML: Benchmarking intelligent systems using LLM Agents, RL (DQN), and Computer Vision.
  • ☁️ Backend Architecture: Deploying microservices with FastAPI & Spring Boot on AWS/GCP.
  • 🌍 Bilingual Engineer: Fluent in English and Japanese; experienced in cross-cultural technical environments.

🚀 Featured Projects

Project Description Stack
GenAI vs RL Tetris AI Agent Benchmarking. Custom Gymnasium env comparing Gemini & OpenAI Agents vs. Deep Q-Networks (DQN). Implemented Chain-of-Thought (CoT) to improve survival rates by 8.2% over baseline. Python Gemini OpenAI
FafnirDT High-Performance DB. Enhanced 2PLSF concurrency protocol. Achieved 610K TPS under delayed workloads through dynamic timestamp injection. C++ Linux Systems
MediSkinAI Medical Computer Vision. Melanoma screening web app (91% accuracy) using a FastAPI backend for model inference on Cloud Run. FastAPI Docker React
Right Source Auto Production Freelance Platform. Migrated dealership from Facebook to custom platform. Built React frontend + Headless Sanity CMS backend. TypeScript Sanity

Pinned Loading

  1. Tetris-Benchmark Tetris-Benchmark Public

    Benchmarking LLM Agents (Gemini & OpenAI) against Deep Q-Networks (DQN) and other RL models in a custom Gymnasium environment. Measures survival rates and decision latency.

    Python 1 1

  2. FafnirDT FafnirDT Public

    FafnirDT extends the 2PLSF (Two-Phase Locking with Starvation Freedom) protocol to fix a critical scalability flaw in lock-based concurrency.

    C++ 1

  3. MediSkinAI MediSkinAI Public

    Melanoma detection web app (91% accuracy) using PyTorch, FastAPI, Langgraph and Docker.

    Python 1

  4. leochoo/maigo leochoo/maigo Public

    A GeoGuessr Clone for MMO

    Svelte 5