Skip to content
View karen1522158173's full-sized avatar
  • Beijing University of Chemical Technology
  • Beijing

Block or report karen1522158173

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 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
Karen1522158173/README.md

Hi, I'm Kailun Liu

Master's student in Information and Communication Engineering at Beijing University of Chemical Technology.
I work on LLM applications, AI Agent systems, RAG workflows, model deployment, and remote sensing intelligence.

About Me

  • Research interests: AI Agent workflows, large language model applications, multimodal understanding, SAR image processing, and remote sensing image registration.
  • Engineering focus: Python/C++ algorithm development, FastAPI services, computer vision pipelines, model inference, edge deployment, and full-stack AI demos.
  • Recent experience: built and deployed an aerospace target intelligent imaging and recognition system during an internship at the Aerospace Information Research Institute, Chinese Academy of Sciences.
  • Current work: building LiquiScope-A, an A-share market liquidity research Agent with tool calling, multi-role analysis, risk review, and report generation.
  • Contact: 16699178595@163.com

Featured Projects

Project What It Shows Stack
Salty-LLM-Platform A full-stack Chinese LLM training and deployment platform, covering TinyLM implementation, Pretrain, SFT, DPO, GRPO, FastAPI inference, and a React web console. PyTorch, FastAPI, React, SQLite, LLMOps
Hello Agent An AI Agent learning and experimentation project from basic LLM calls and tool registration to ReAct, Plan-and-Solve, LangGraph, AutoGen, and AgentScope multi-agent examples. Python, ReAct, LangGraph, AutoGen, AgentScope
LiquiScope-A An A-share market liquidity research Agent prototype with market data tools, liquidity indicators, multi-role analyst nodes, risk review, human-in-the-loop confirmation, and research report generation. Python, LangGraph, FastAPI, Pandas, DuckDB, Tushare, AKShare

Tech Stack

LLM / Agent
Prompt Engineering, RAG, Embedding Retrieval, Tool Calling, ReAct, CrewAI, LangGraph, AutoGen, AgentScope, Human-in-the-loop

Deep Learning / CV
PyTorch, Transformers, PEFT/LoRA, OpenCV, YOLOv5, SAR Image Processing, Image Registration, Object Detection

Backend / Deployment
FastAPI, HTTP API, SFTP, Socket, Linux/Ubuntu, Docker, CUDA, cuDNN, TensorRT, Jetson AGX Orin

Data / Research Tools
Pandas, DuckDB, NumPy, SciPy, Tushare, AKShare, BaoStock, Vector Search, Structured Report Generation

Frontend / Applications
React, Vite, PyQt5, Flask, Streamlit-style AI application prototyping

Experience

  • Aerospace Information Research Institute, Chinese Academy of Sciences - AI Application R&D Intern - 2025.10 - 2026.01
    Built and integrated an aerospace target intelligent imaging and recognition system on NVIDIA Jetson AGX Orin, connecting SAR echo simulation, SAR imaging, YOLOv5 object detection, result export, and PC-edge device communication.

Education

  • Beijing University of Chemical Technology - Information and Communication Engineering - M.S. candidate - 2024.09 - 2027.06
  • Southwest University of Science and Technology - Information Countermeasure Technology - B.Eng. - 2019.09 - 2023.06

Research

  • MCDCA-Net: hybrid convolution and dynamic channel attention network for SAR image registration.
  • SAMR-Net: semantic-guided adaptive multiscale registration network for optical-SAR image registration.
  • BACF-UNet: improved UNet for vehicle detection in SAR images.
  • Change-guided difference interaction attention network for remote sensing change detection.

GitHub Focus

I keep this profile focused on reproducible AI engineering projects.
Some research code is not public yet because related papers and experiments are still in progress.

Pinned Loading

  1. MSNR MSNR Public

    Python 1

  2. Salty-LLM-Platform Salty-LLM-Platform Public

    "Salty LLM Platform: a full-stack Chinese LLM project from scratch — Transformer, pre-training, SFT, DPO, GRPO, inference, and web console. Showcases LLM training & LLMOps, not just API calls."

    Python

  3. agent agent Public

    "An AI Agent learning project for Chinese developers. Covers LLM calls, tool registration, ReAct loops, then advances to Plan-and-Solve, LangGraph, AutoGen, and AgentScope multi-agent practices. Gr…

    Python