Skip to content
View Ailta666508's full-sized avatar
2 followers · 4 following

Block or report Ailta666508

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.

Content in all repositories owned by your account will be closed.
Maximum 250 characters. Please don’t include any personal information such as legal names or email addresses. Markdown is supported. This note will only be visible to you.
Report abuse

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

Report abuse
Ailta666508/README.md

Zihan Shen

I build AI systems that connect research ideas with working software: agent workflows, graph learning and reinforcement learning, generative modeling, simulation, and full-stack AI products.

Selected projects

Project What it demonstrates
AutoPersona April–July 2026 research project on structured personalized memory, proactive clarification, SFT/GRPO, and node-level agent execution; the repository publishes a tested memory/runtime reference implementation.
PPO-STGNN Graph-based reinforcement learning for DAG task scheduling across cloud, edge, and end devices; research code for an ICSINC 2026 paper.
Hard-Minor-GAN Boundary-focused GAN augmentation for imbalanced tabular classification, with feature selection and XGBoost evaluation. Outstanding 2025–2026 fall Data Mining course project.
Puppy Journey Showcase An AI-native full-stack product built with Next.js, TypeScript, LangGraph, multimodal generation, Supabase, and production-oriented state and data flows.
neon-runner Outstanding 2025–2026 The Practice of AI Applications in Games course project: a cyberpunk stealth game with FSM drones, A* pathfinding, and an LLM commander.
HSIR Adaptive Hypergraph A multi-risk-aware adaptive hypergraph model with Monte Carlo simulation for spreading dynamics. Outstanding AI for Complex Networks course project.
Xiaoxiang Store Analytics End-to-end retail analytics: demand-driver analysis, five-band XGBoost forecasting, site-selection support, and a Flask dashboard.

Areas of interest

  • Agent systems, memory, evaluation, and multimodal interaction
  • Graph neural networks, reinforcement learning, and scheduling
  • Generative modeling and learning from imbalanced data
  • Reproducible experimentation and full-stack AI engineering

Technical toolkit

Languages: Python, TypeScript, JavaScript, SQL
AI and data: PyTorch, TensorFlow, scikit-learn, XGBoost, LangGraph
Product engineering: Next.js, React, Supabase, Flask, Docker

Benchmarking work

My benchmark and evaluation repositories are kept separate from the selected original projects above: Persona2Web benchmark, Persona2Web evaluation, PI benchmark, and PI-Bench reproduction.

Pinned Loading

  1. AutoPersona AutoPersona Public

    Structured personalized memory and proactive clarification for long-horizon LLM agents — research project and reference runtime (Apr–Jul 2026).

    Python 1

  2. PPO-STGNN PPO-STGNN Public

    Research code for PPO-STGNN: graph-based reinforcement learning for DAG scheduling in cloud-edge-end systems (ICSINC 2026).

    Python 1

  3. Hard-Minor-GAN Hard-Minor-GAN Public

    Boundary-focused GAN augmentation for imbalanced customer-churn prediction — outstanding 2025–2026 fall Data Mining course project.

    Python

  4. puppy_journey_showcase puppy_journey_showcase Public

    AI-native full-stack app with LangGraph orchestration, multimodal generation, Supabase data flows, and a production-oriented Next.js architecture.

    TypeScript

  5. neon-runner neon-runner Public

    Outstanding 2025–2026 The Practice of AI Applications in Games course project: a cyberpunk stealth game with FSM drones, A* pathfinding, and an LLM commander.

    JavaScript

  6. HSIR-Adaptive-Hypergraph HSIR-Adaptive-Hypergraph Public

    Multi-risk-aware adaptive link breaking and spreading dynamics on hypergraphs — outstanding AI for Complex Networks course project.

    Python 1