Skip to content
View hcy1041116's full-sized avatar

Block or report hcy1041116

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
hcy1041116/README.md

你好,我是 Kelsey 👋

正在轉職 資料工程師 / AI 應用工程師,專注用 Python 打造穩定的資料管線與實用的 AI 產品。

🔭 目前持續深化資料工程(Airflow、Docker、SQL)與 LLM 應用開發(RAG、agent、FastAPI)。


🛠️ 技術棧

  • 語言: Python · SQL · TypeScript
  • 資料工程: Apache Airflow · Pandas · MySQL · ETL / star-schema 建模
  • 後端與 AI: FastAPI · LangChain · RAG · ChromaDB
  • 工具: Docker · Git

🚀 主要專案

Documind 多 PDF 問答助手,基於 RAG(FastAPI + LangChain + ChromaDB + PostgreSQL),具備對話記憶並支援雲端與本地模型 (GPT-4o / Ollama) 切換。 為了讓系統「可被量化迭代」,自建了 evaluation harness:標注測試集、retrieval 指標 (Recall@k / MRR) 與 LLM-as-judge faithfulness 評分,並將 prompt 抽離程式碼以利 A/B 比較。 目前正升級為 agentic RAG(LangGraph):query rewriting、reranker 與自我校正檢索,並導入 LangFuse 做全流程追蹤。

ai-classroom 全端學習平台(團隊專案),負責後端 API 設計,以及一條將非結構化文章轉為訓練情境的資料處理 pipeline(清整 → 多階段 LLM 生成 → 品質檢核)。

airflow-etl 以 Apache Airflow 編排的資料管線。


English

Hi, I'm Kelsey 👋

Aspiring Data Engineer / AI Application Engineer focused on building reliable data pipelines and practical AI products with Python.

🔭 Currently deepening data engineering (Airflow, Docker, SQL) and LLM application development (RAG, agents, FastAPI).


🛠️ Tech Stack

  • Languages: Python · SQL · TypeScript
  • Data Engineering: Apache Airflow · Pandas · MySQL · ETL / star-schema modeling
  • Backend & AI: FastAPI · LangChain · RAG · ChromaDB
  • Tooling: Docker · Git

🚀 Featured Projects

Documind RAG-based multi-PDF assistant (FastAPI + LangChain + ChromaDB + PostgreSQL) with conversation memory and switchable cloud/local LLMs (GPT-4o / Ollama). Built an evaluation harness to make the system measurably iterable: a labeled test set, retrieval metrics (Recall@k / MRR), and LLM-as-judge faithfulness scoring, with prompts decoupled from code for easy A/B comparison. Currently upgrading to an agentic RAG system (LangGraph) with query rewriting, reranking, and self-correcting retrieval, plus LangFuse for end-to-end tracing.

ai-classroom Full-stack learning platform (team project). Designed backend APIs and built a data pipeline transforming unstructured articles into training scenarios (cleaning → multi-stage LLM generation → quality review).

airflow-etl Orchestrated data pipeline built with Apache Airflow.

Popular repositories Loading

  1. SAS-txt-multiple-import SAS-txt-multiple-import Public

    This project can import multiple txt files into SAS, using SAS and python.

  2. vlm-annotation-qc vlm-annotation-qc Public

    Python

  3. Documind Documind Public

    Python

  4. ai-classroom ai-classroom Public

    TypeScript

  5. order-star-etl order-star-etl Public

    Python

  6. airflow-etl airflow-etl Public

    Python